Electric Power Systems Research 231 (2024) 110270

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Enhancing voltage profile in islanded microgrids through hierarchical

control strategies
Sayed Hamid Hosseini Dolatabadi a, Alireza Soleimani b, Afshin Ebtia c, Miadreza Shafie-khah d, *,
Tanveer Hossain Bhuiyan a, *
a
Department of Mechanical Engineering, The University of Texas at San Antonio, San Antonio, TX 78249, USA
b
Department of Energy Engineering, Sharif University of Technology, Tehran 11365-11155, Iran
c
Concordia Institute for Information System Engineering, Concordia University Montreal, 455 Boul. de Maisonneuve Ouest, Montréal, QC H3G 1M8, Canada
d
School of Technology and Innovations, University of Vaasa, Vaasa, Finland

A R T I C L E I N F O A B S T R A C T

Keywords: This paper presents a novel approach aimed at enhancing the voltage profile of an islanded microgrid (MG)
Islanded microgrid through the utilization of Phasor Measurement Units (PMUs) and a hierarchical control system for Distributed
Tertiary control Generators (DGs). Recognizing the criticality of maintaining an optimal voltage profile for essential loads within
Hierarchical microgrid structure
an MG, we assume that PMUs are strategically placed across the MG based on existing placement methodologies
Droop control
Wams
to ensure the observability of load buses. The synchrophasor measurements captured by these PMUs are
leveraged within an offline-designed tertiary control layer of the hierarchical framework, which in turn sets the
voltage reference for the secondary control layer of DGs across the MG. This approach effectively enhances the
voltage profile of vital load buses within the MG. The proposed hierarchical strategy encompasses primary,
secondary, and tertiary levels of control, responsible for stabilizing voltage and frequency, computing power
references, and managing power flow, respectively. PMU measurements play a crucial role in the secondary
control level by continuously monitoring deviations. By ensuring the effective placement of PMUs, the proposed
framework enables efficient control, leading to notable improvements in the voltage profile and power man­
agement. To evaluate the effectiveness of our approach, five test networks were used: 14-bus, 30-bus, 39-bus, 57-
bus, and 118-bus test networks. Simulation results demonstrate an impressive over 80 % enhancement in the
MG’s voltage profile achieved by our proposed method. Additionally, our heuristic strategy efficiently assigns the
closest load bus to each DG while aggregating equidistant bus deviations. Collectively, our research underscores
the significant impact of PMU measurements and a well-structured hierarchical voltage control system in sub­
stantially improving the voltage profile of MGs.

to optimize system operation [4].

A significant amount of research has been conducted on the optimal
1. Introduction PMU placement (OPP) problem considering different aspects of the
problem including but not limited to fundamental state observability
MGs utilize DGs and storage to supply dependable energy to specific analysis such as numerical and topological observability analysis [5],
locations, operating either in grid-connected or islanded mode [1]. So­ addressing practical constraints like N-1 contingencies [6], the exten­
phisticated hierarchical control strategies are necessitated to ensure sion of the OPP problem to consider various PMU applications such as
stability and reliability [2]. PMUs enhance MG monitoring and fault location [7,8], and development of exact methods to solve the
controllability by providing synchrophasor data for real-time situational linear mathematical formulation of the OPP problem [9]. The optimal
awareness. This data is leveraged in numerous applications like fault placement of PMUs and the observability analysis discussed in the
and islanding detection, and system restoration [3]. Specifically, PMU literature have significant implications for the hierarchical control of
voltage, frequency, and phase angle measurements enable detecting MGs. By leveraging the real-time measurements provided by PMUs,
deviations from optimal voltage profiles, allowing secondary controllers

* Corresponding authors.
E-mail addresses: sayedhamid.hosseinidolatabadi@utsa.edu (S.H.H. Dolatabadi), alireza.soleimani@alum.sharif.edu (A. Soleimani), afshin.ebtia.0995474@
concordia.ca (A. Ebtia), mshafiek@uwasa.fi (M. Shafie-khah), Tanveer.bhuiyan@utsa.edu (T.H. Bhuiyan).

https://doi.org/10.1016/j.epsr.2024.110270
Received 26 November 2023; Received in revised form 29 January 2024; Accepted 24 February 2024
Available online 8 March 2024
0378-7796/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Nomenclature ω∗MG Reference MG frequency

ΔE(S) Laplace transformation of voltage compensation
kP Proportional control gain V
̂i Voltage phasor at bus i
kI Integral control gain ̂
θj jth measurement obtained from PMUs
Vi Voltage phasor of bus i obtained by PMUs H The measurement Jacobian matrix
V rated Rated voltage phasor of MG MG Microgrid
ΔVi Deviation of voltage phasor of bus i from the rated voltage PMU Phasor Measurement Units
phasor DG Distributed Generators
δVi Output of PI controller in islanded tertiary control WAMS Wide Area Management System
E∗i Output of islanded tertiary control for bus i as the reference OPP Optimal PMU placement
of secondary control VSG Virtual Synchronous Generators
EMG Actual reference MG voltage magnitude VSC Voltage source converter
E∗MG Reference MG voltage magnitude PI Proportional-integral
ωMG Actual MG frequency

Fig. 1. General perspective of the focus area of this research.

advanced control strategies can be implemented to enhance the opera­ MGs, ensuring accurate load sharing without requiring changes for
tion and stability of MGs. added/removed DGs or loads. España et al. [16] discussed a distributed
Shan et al. [10] designed a fuzzy controller for improving the voltage optimization technique for optimal active and reactive power flow in
quality of distributed generators in islanded mode, outperforming con­ islanded MGs. Alghamdi et al. [17] presented a decentralized secondary
ventional methods in simulations. Moutevelis et al. [11] simulated a control strategy to mitigate voltage harmonics in isolated MGs using a
secondary controller for distributed generators to regulate voltage in modified droop controller and harmonic droop controller. Hou et al.
distribution networks, which periodically updates virtual admittance [18] used a decentralized control approach for an isolated MG using
gains through optimization to prevent voltage fluctuations. Abouassy, consensus algorithms. Hernández et al. [19] designed a vehicle-to-grid
Ahmed M., et al. [12] discussed the power-sharing problem in islanded model to simultaneously provide primary frequency control and dy­
MGs and presented various droop-based control techniques to address namic grid support at the plug-in terminal, without disturbing scheduled
this issue. Hui et al. [13] presented a coordination control method for charging. Bueno et al. [20] assessed the stability of transmission systems
isolated urban MGs with highly distributed generator penetration. The with large utility-scale PV plants during various disturbances using
method taps flexibility from both supply and demand sides to improve quantitative stability indices. Askarzadeh et al. [21] analyzed voltage
stability and resilience. Meng et al. [14] modeled a two-layer frequency behavior in distribution systems considering uncertainties from PV and
control strategy for MGs, with lower–layer control integrating primary electric vehicle charging loads using a probabilistic approach. Feng et al.
and secondary control and lower–layer optimizing power sharing to [22] discussed an enhanced MG power flow approach that incorporates
optimize economic operation at a short timescale. Mi et al. [15] simu­ hierarchical control effects for accurate power sharing and voltage
lated a droop control for parallel distributed generators in islanded DC regulation. Golsorkhi et al. [23] used a decentralized control approach

2
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Fig. 2. Correlation of three levels of hierarchical control strategy for an MG.

for an isolated MG using consensus algorithms. Droop control is used at To provide a visual overview of the key contributions, Fig. 1 illus­
the DER level for load sharing, and a correction voltage is added to trates the focus areas of this research. Fig. 1. (a) shows the general
eliminate the effect of line impedances on current sharing accuracy. Su landscape, positioning microgrids and inverter-based distributed gen­
et al. [24] utilized a control strategy using synchronized measurements erators as the technologies of interest and the situation of hierarchical
and an adaptive delay compensator to enhance secondary frequency control structure in this specific research area. Zooming in, Fig. 1. (b)
control in islanded MGs, validated through numerical simulations. As depicts the hierarchical control framework, which provides a high-level
recent studies have [25,26] control schemes allow emulating the layout of the techniques developed in this paper and their imple­
behavior of synchronous generators through power electronic con­ mentation within microgrid hierarchical control.
verters. This facilitates provision of grid support functionality and Moreover, Fig. 2 presents a comprehensive overview of the hierar­
inertia emulation to enhance frequency regulation. In the context of chical control strategy in microgrids, highlighting the integration of our
hierarchical microgrid control, VSG algorithms could be integrated at proposed tertiary control strategy within the framework. The details of
the primary level to enable grid-forming capabilities and inertial the tertiary control strategy and its impact on voltage regulation are
response. The secondary level could potentially override VSG setpoints elaborated in the subsequent sections.
based on optimization and tertiary objectives. While the focus of the To propel the investigation in this specific research domain, this
current work is the coordination of secondary and tertiary control, paper presents a novel hierarchical control approach to enhance the
analyzing the integration of VSG-based primary control merits explo­ voltage profile of an islanded MG utilizing PMU measurements. Strate­
ration in future work. VSG schemes represent a promising approach to gically placed PMUs provide observability of load buses. A tertiary
facilitate stable operation and grid support within islanded microgrids, control layer is introduced that integrates these measurements to
and their application warrants consideration within the hierarchical calculate voltage deviations, providing reference signals to the under­
coordination framework proposed here. lying secondary control of DGs to minimize deviations. Therefore, the

3
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

key contributions of this study are as follows: regulated based on predetermined reference values. This control strat­
egy is executed using a voltage source converter (VSC) with current
1) We propose a novel tertiary control strategy that leverages PMU data control. In voltage control mode, the DG functions as a VSC, and the
to improve the voltage profile of islanded microgrids within the hi­ primary control sets the reference voltage. Typically, this reference
erarchical control framework. This integrated approach allows the voltage is determined using traditional droop characteristics [27].
tertiary level to optimize voltage regulation using real-time phasor
measurements. 2.2. Secondary control
2) We develop an efficient offline design methodology for the tertiary
control layer based on network topology and location of DGs. This The secondary control serves the objective of mitigating the voltage
provides a computationally simple way to determine voltage set­ and frequency deviations within the MG, and is responsible for
points for secondary control without complex real-time computing the active and reactive power references, which are then
optimization. transmitted to each power converter. These values allow for the
3) By leveraging PMU infrastructure to coordinate hierarchical control, adjustment of the droop control slope in the primary control. The extent
our findings demonstrate the immense potential to effectively tackle of frequency deviation is contingent on the active power delivered to the
the key challenge of voltage management in islanded microgrids. connected load. Hence, the balance between the connected load and
The analysis provides valuable insights into the design trade-offs generated active power can serve as a basis for determining control
between different tertiary control architectures using practical case parameters aimed at offsetting this frequency variation [28]. A
studies. Our findings highlight the immense potential of leveraging proportional-integral (PI) controller can be employed for frequency
PMU infrastructure to tackle a key challenge of voltage control in control and compensation as Eq. (1).
islanded microgrids through hierarchical coordination resulting in a ( )
kI ( ∗ )
substantial improvement in the voltage profile of islanded Δω(S) = kP + ωMG − ωMG + (Δωss ) (1)
S
microgrids.
Where kP and kI are the control gains. Also, ωMG and ω∗MG are the actual
2. Hierarchical control of MGs
and reference MG frequency. The error between ωMG and ω∗MG is miti­
gated through the utilization of a PI controller in conjunction with Δωss
The MG’s control strategy involves primary, secondary, and tertiary
which is the frequency derived from the synchronization loop of the
control levels addressing varying timeframes necessitating a hierarchi­
utility grid and will be omitted in the islanded mode. The frequency
cal control structure that caters to each requirement at separate control
control parameter can be incorporated into the droop control of the
hierarchies. The primary control in MGs stabilizes voltage and fre­
power converter to address frequency errors.
quency, distributes generation power among DGs, and mitigates circu­
Taking the same argument into account, voltage magnitude
lating current to prevent equipment damage. However, the primary
compensation can also be achieved in the secondary control using a PI
control has the potential to induce frequency and voltage variation even
controller. PI control parameters are calculated for each converter
during stable conditions. The secondary control corrects voltage and
connection point. The voltage controller in the secondary control uses
frequency issues in the MG caused by the primary control and operates
the voltage profile of each converter to determine deviations and com­
slower than the primary control to maintain separate control loops. This
pensates for them through voltage adjustments in the primary control.
decoupling enables separate design considerations for each control loop
This ensures that every converter can simultaneously engage in voltage
and facilitates their individual optimization. The tertiary control as­
compensation. The voltage magnitude compensation signal can be
sumes responsibility for managing power flow between the MG and the
calculated as Eq. (2).
main grid, with a primary focus on achieving optimal economic opera­
( )
tion. This control level operates at a slower pace compared to the pre­ kI ( ∗ )
ΔE(S) = kP + EMG − EMG (2)
ceding levels and takes economic considerations into account when S
determining the most efficient functioning of the MG. When connected
to the main grid, DGs can regulate the power flow between the MG and Where kP and kI are the control gains. Also, EMG and E∗MG are the actual
the main grid by adjusting the voltage magnitude and frequency. and reference MG voltage magnitude, and ΔE(S) is the Laplace trans­
However, in an islanded MG that has no energy transaction with the formation of voltage compensation parameter which is transmitted to
grid, stabilizing the voltage profile of the MG is highly important that the primary control of the converter.
can be handled by a new tertiary control structure proposed in this
paper. 2.3. Tertiary control
In the proposed methodology, synchrophasor measurements play a
crucial role in capturing the dynamic behavior of the MG. Fig. 2 shows The main objective of tertiary control in grid-connected mode is to
the hierarchical control strategy for an MG in which the three-level determine the operational level of active power for each DG, enabling
control mechanisms communicate with each other through communi­ power flow control between MG and the grid. By adjusting the frequency
cation systems. It should be noted that in the islanded mode, the purpose and magnitude of DGs voltages, the power flow can be regulated
of implementing a tertiary control is to provide voltage magnitude and effectively. However, in the islanded mode, the tertiary control should
frequency references for secondary control of DGs to improve the ensure that MG operates within its rated limits. Hence, it should provide
voltage profile of the system. Since the focus of this paper is studying the the frequency and voltage magnitude references for DGs to ensure that
MG operation in islanded mode, a novel tertiary control mechanism is MG can satisfy the frequency requirements and also compensates for the
proposed, shown as islanded tertiary control in the figure, to enhance voltage deviation of the critical loads. In this regard, the details of the
the voltage of buses supplying critical loads. The details of each level of control mechanism for tertiary control of islanded MG are proposed in
MG hierarchical control strategy are explained as follows [3]. section 3.

2.1. Primary control 3. Proposed control strategy

The primary control sets the reference values for voltage and current In this section, a novel framework is proposed to use the capability of
control loops of DGs, which can operate in either PQ or voltage control synchrophasors provided by PMUs to improve the voltage profile of the
mode. In PQ mode, the DGs’ active and reactive power delivery is MG based on the hierarchical structure of the control framework of DGs.

4
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Fig. 3. The proposed closed-loop control structure for tertiary control in islanded mode.

Fig. 4. Schematic of the control block for voltage reference in islanded tertiary control.

As explained in Section 2, the secondary control is responsible for control structure is proposed based on the available voltage phasors of
mitigating the frequency and voltage deviations. This is achievable load buses and the secondary control of DGs to improve the voltage
through appropriate settings of voltage and frequency reference in the profile of MG. The proposed closed-loop control structure is shown in
droop control framework of the secondary control level. Choosing a Fig. 3.
proper reference can help MG to improve the voltage profile of critical In the proposed framework, the voltage profile is calculated from
loads. Since we assumed that PMUs are strategically placed within the phasor measurements obtained through PMUs using state estimation
MG ensuring observability of important buses, voltage phasor mea­ techniques. The general state estimation process can be mathematically
surements of all load buses can be captured. These voltage phasor described as Eq. (3):
measurements, obtained through the synchrophasors provided by ∑
PMUs, serve as valuable real-time data for monitoring voltage deviations
̂ i = f (̂
V θ) ≅ Hij × ̂
θj (3)
and formulating appropriate control actions. It is essential to highlight
j

that our proposed framework, while not directly focused on PMU

Where V ̂ i represents the estimated voltage phasor at bus i, ̂θ j is the jth
placement, acknowledges and accommodates uncertainties within the
measurement obtained from PMUs, and H is the measurement Jacobian
microgrid system. The strategic placement of PMUs, as assumed in our
matrix. Moreover, in the context of microgrids, achieving an accurate
model, takes into account potential contingencies such as load fore­
voltage profile through state estimation involves considerations distinct
casting errors and component outages. We recognize that uncertainties
from conventional power systems. Unlike conventional power systems,
in PMU placement strategy and state estimation can arise from various
microgrids exhibit unique operational characteristics, necessitating a
factors, including loss of communication infrastructure, outage contin­
tailored approach to state estimation. Notable differences arise from the
gencies of lines, DGs, PMUs, and current measurement channels of
inclusion of frequency as a state variable, the implementation of droop
PMUs. Our approach not only assumes complete state observability but
control in primary regulation, and the existence of secondary and ter­
also accounts for uncertainties in PMU placement, allowing the system
tiary regulation mechanisms. In microgrids, the frequency, in addition
to adapt to unforeseen variations. The robustness of our design is further
to voltage, plays a pivotal role in system dynamics, requiring specialized
emphasized by the offline design methodology based on network to­
algorithms to account for its influence as a state variable. Furthermore,
pology, providing inherent resilience without requiring precise oper­
the incorporation of droop control, inherent in primary regulation, in­
ating point estimates. The hierarchical control strategies, incorporating
troduces nonlinearities that demand specific treatment in the state
adaptive mechanisms and real-time feedback, dynamically respond to
estimation process. The coexistence of secondary and tertiary regulation
changes in load conditions, effectively mitigating the impact of un­
layers in microgrids further distinguishes the state estimation frame­
certainties in load forecasting and outages. By leveraging the availability
work, as these layers contribute to the overall hierarchical control
of voltage phasors at load buses, our control strategy utilizes these
strategies. While conventional state estimation techniques form the
measurements in the tertiary control layer of the hierarchical framework
basis, adapting and extending them to accommodate these microgrid-
to improve the voltage profile of the MG. Therefore, a closed-loop
specific elements is crucial for accurately capturing the system’s state

5
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Table 2
Table 1
Voltage deviations of load buses with respect to the single-
General parameters of the modified standard test networks.
feedback scenario.
Parameter Rated value/Description
Bus Number Voltage Deviation (V)
rated
V 220V
frated 50Hz 1 -12.01
Srated 100VA 2 -9.674
Zbase 48.4Ω 3 -5.614
DGs 3-phase PWM-based inverters 4 -6.427
DGs inner loop control KV 0.005p.u. 5 -7.317
6 -5.279
Kf 0.02p.u.
9 2.746
10 3.071
11 -0.1858
and ensuring effective control [29].
12 -1.529
Therefore, phasor measurements provided by PMUs are gone 13 -0.06243
through the phasor state estimator and the voltage phasors of all critical 14 5.528
buses are estimated. It is important to note that the detailed formulation
of the state estimation process and its associated matrices are derived
from the methodology presented [29,30] addresses static state estima­ Table 3
tion in electric power systems. Our work builds upon this established Voltage deviations of load buses with respect to the five in­
methodology, assuming its application for obtaining accurate voltage dependent feedbacks scenario.
profiles in the microgrid context. [30]. Then, based on the difference Bus Number Voltage Deviation (V)
between the desired voltage magnitude and the real-time voltage
1 -0.543
magnitude of all critical buses, appropriate reference points for the 2 -0.9408
secondary control loop of all DGs across the MG are provided. The same 3 -10.82
approach is chosen for the frequency control. The voltage deviations are 4 -3.618
gone through a PI controller to minimize the error in the closed-loop 5 -2.837
6 -4.818
feedback control. The control block uses the output of the PI control­
9 0.8887
lers as the input signals to provide better set-points for the voltage 10 1.64
reference of the secondary control level of DGs across the MG. The 11 -0.6801
general schematic of the control block is shown in Fig. 4. 12 -1.241
As can be seen, the proposed control block gathers voltage deviation 13 0.07759
14 4.563
data across the MG to provide improved secondary control setpoints for
DGs. A controller should find an appropriate relationship between the
inputs and the outputs of the control block that can be designed in two
ways: (1) Real-time optimization-based design, and (2) Offline design. Table 4
Voltage deviations of load buses with respect to the combi­
Real-time optimization utilizes PMU measurement to minimize
natorial feedbacks scenario.
voltage deviations via an optimization-based simulation to obtain
optimal references for the secondary control loops to minimize the total Bus Number Voltage Deviation (V)

voltage deviation across the network. However, due to the nonlinear and 1 -1.103
simulation-based nature of the real-time optimization problem, it can be 2 0.2284
3 4.305
computationally extensive. In contrast, the offline design uses some
4 -0.527
heuristic methods to obtain a proper combination of voltage deviations 5 -0.7151
for each of the DGs. Although the offline design cannot guarantee the 6 -4.384
minimum voltage deviation across the network, it is definitely less time- 9 -2.62
consuming and computationally less extensive. Therefore, in this paper, 10 -1.149
11 -1.878
based on a comprehensive analysis of different combinations of voltage
12 -1.093
deviations discussed in section 4, a heuristic approach to obtain the best 13 -0.034
offline design for the control block is proposed to improve the voltage 14 2.577
profile of the MG in an islanded mode. As this control strategy is based

Fig. 5. Three proposed control structures for the IEEE 14-bus network. a) Single-feedback, b) five independent feedbacks, and c) combinatorial feedbacks.

6
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Fig. 6. Comparing three control strategies.

on the network topology, it is impossible to propose a general structure • Existing generators were modeled as distributed inverter sources to
for the architecture of the control block. However, some rules to emulate the power electronics-interfaced sources prevalent in
improve the design of the control block are proposed. microgrid systems.
• Network impedances were proportionally scaled down to accurately
1- The voltage deviation of all observable load buses should be included reflect the lower 220 V distribution voltages according to ZΩ = Zp.u. ×
in the design. Zbase , where Zp.u. is the per-unit impedance of the standard test net­
2- For each of the DGs, assign the closest load bus for the closed-loop works. Moreover, the demands are proportionally scaled down to the
feedback. rated power of 100VA.
3- If a load bus has almost the same topological position to more than
one DG, assign the voltage deviation of this bus to all the DGs with The general parameters of the modified test networks are described
almost the same distance. in Table 1.
4- The remaining DGs receive feedback from the remaining buses with The test networks were transitioned to islanded operating mode and
the least distance. integrated with the hierarchical control framework needed to evaluate
5- If more than one load buses have almost the same topological dis­ our tertiary control approach for standalone microgrids. Through these
tance from a DG, assign the aggregated voltage deviations of those adjustments to translate the original test cases into representative
load buses to that DG. microgrid models, we aimed to provide standardized, validated bench­
marks to assess the performance of the proposed voltage control meth­
The above rules are the main principles of designing an appropriate odology. The scaled voltage levels, distributed generation sources,
control strategy for the control block shown in Figs. 3 and 4. reduced impedances, and implementation of islanded hierarchical con­
trol provide appropriate simulation environments to demonstrate the
4. Numerical results improvements enabled by our tertiary control contribution.

To assess the efficiency of our proposed islanded tertiary MG control, 4.1. Offline tertiary control design for IEEE 14-Bus network
our methods are applied to five different test networks: IEEE 14-Bus
[31], IEEE 30-Bus [32], New-England 39-Bus [33], IEEE 57-Bus [34], In this section, some control designs for the control block shown in
and IEEE 118-Bus [35] standard test networks. In the proposed hierar­ Figs. 3 and 4 are proposed to analyze the performance of the system and
chical control methodology, the values of kp and kl are control gains that the voltage profile of critical loads. We designed three scenarios to
should be determined through appropriate design or tuning methods. In explain and evaluate the effectiveness of our proposed control strategy.
our simulation, the parameters are considered as: kp = 1, kl = 2 for Eq. To simulate the scenarios, the IEEE 14-Bus network with the necessary
(1) and kp = 1, kl = 4 for Eq. (2). modifications to enable hierarchical control is considered.
Simulink environment of MATLAB is used for simulating the MG and
the proposed hierarchical control framework. The computational time 4.1.1. Single feedback scenario
for simulating the hierarchical control of MG in Simulink was from 65 s To enable a comparative analysis, we implemented a single-feedback
for IEEE 14-Bus test network to 15 min for IEEE 118-Bus test network on control strategy as a representative approach of existing techniques that
a core i-9, 2.5 GHz processor computer. use a single voltage reference for distributed generators across the
To adapt the standard IEEE test networks as representative islanded microgrid. Due to the lack of prior studies directly comparable to our
microgrid models for evaluating our proposed control strategy, we made proposed tertiary control design and specific focus on voltage profile
the following modifications: enhancement, this single-feedback scenario provides a baseline for
comparison to demonstrate the relative improvements enabled by our
• The network voltage levels were reduced to a distribution level of approach.
220 V to be representative of typical microgrid voltage standards. In the first scenario, the control block only uses the voltage phasor of

7
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Fig. 7. The schematic of control block for tertiary control level of a) 30-bus network, and b) 39-bus test network.

bus 13 to produce reference settings for the secondary control of all DGs. from bus 13 from the topological point of view. Therefore, to minimize
The control strategy in this scenario is shown in Fig. 5. a. the voltage deviation of bus 13 which is a load bus that has a voltage
Table 2 displays the voltage deviations of load buses for this lower than the rated voltage, bus 1 needs to increase its voltage more
particular control strategy. than other DGs. The total absolute voltage deviation of all load buses in
As can be seen from Table 2, the minimum and the maximum voltage this scenario is 59.45v which is 0.27pu at the 220v voltage level of the
deviations are for buses 13, and 1, respectively. Due to having the single system. In other words, the total absolute voltage deviation in the single-
feedback from the voltage phasor of bus 13, it was definitely expected feedback scenario is 27 %.
that the entire secondary systems are working together to minimize the
voltage deviation of bus 13 and it should have the minimum voltage 4.1.2. Five independent feedbacks scenario
deviation across the MG. On the other hand, bus 1 is the furthest bus In this scenario, the control block uses the voltage phasors of five

8
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

Fig. 8. Comparing total absolute voltage deviation for test networks with three control strategies.

references of DGs in buses 2 and 3. The voltage deviations for this sce­
Table 5 nario are shown in Table 4.
Results of applying the proposed islanded tertiary control for test networks. Fig. 6 compares the three scenarios studied in this section regarding
Test Network Run-time (Seconds) Voltage Profile Improvement the average, maximum, and total absolute value of the voltage devia­
IEEE 30-Bus 173.2 80 % tion. The maximum deviation for the combinatorial feedbacks scenario
New-England 39-Bus 210.3 81 % is still for bus 3. However, it is reduced from 10.8v to almost 4.3v which
IEEE 57-Bus 493.6 82 % is a 60 % reduction. The total absolute voltage deviation of all the load
IEEE 118-Bus 886.5 81 %
buses was also reduced to 20.61v which is almost 0.094pu. The proposed
combinatorial control strategy in this scenario improved the voltage
different load buses independently for each of the DGs. In other words, profile of the MG by almost 37 % compared to the second scenario and
the reference of the secondary control loop of each of the DGs will be set by 65 % compared to the first scenario.
by the feedback from the voltage deviation of one of the load buses Therefore, it can be seen that using the available PMU data and the
across the MG. The control strategy for this scenario is shown in Fig. 5. b. proposed tertiary islanded control strategy, the overall voltage profile of
Moreover, Table 3 presents the voltage deviations in the second control the MG is improved and the total absolute voltage deviation of all the
scenario. load buses reduced from 0.27pu to 0.094pu. This improvement was
Table 2 shows a clear improvement in the voltage profile of the MG. achieved without the complexities of real-time optimization and was
The maximum voltage deviation was reduced from 12v for bus 1 in the solely based on the network topology and the position of DGs across the
first scenario to less than 11v for bus 3 in this scenario. Moreover, the MG.
total absolute voltage deviation of all the load buses was reduced to
32.66v which is 0.15pu. This change in the control strategy resulted in a 4.2. Larger test networks results
0.12pu reduction in the total voltage deviation of all the load buses.
Therefore, it can be concluded that this control strategy resulted in a 45 The architecture of the control blocks for 30-Bus and 39-Bus test
% improvement in the voltage profile of the MG. networks based on the proposed heuristic control strategy in section 3,
are shown in Fig. 7. The schematic of control block for other two net­
4.1.3. Combinatorial feedbacks scenario works cannot be shown due to the large scale of the network and the
In the third scenario, the control signals are from the same load buses number of buses.
as in the second scenario. However, the control strategy comes with a Based on the control strategy designed for these test networks, the
small difference based on the proposed rules in section 3. In this sce­ results of the voltage profile of the networks assumed as isolated MGs
nario, the voltage deviation of some of the buses can be fed to the sec­ are shown in Fig. 8 and Table 5. The base voltage for calculating the per-
ondary control loops of different DGs, and also a reference of a DG can be unit value of total absolute voltage deviations is 220V.
a mixture of the voltage deviation of more than one of the load buses The base case scenario in this experiment is to have the rated voltage
across the MG. The control strategy in this scenario based on the as the reference of the secondary control loop of each of the DGs. Table 5
network topology is shown in Fig. 5. c. shows the effectiveness of our proposed heuristic control design based
In IEEE 14-Bus network, buses 14 and 10 have almost the same to­ on some simple rules related to the topology of the MG and the place­
pological distance from the perspective of bus 8 and therefore, it seems ment of DGs. Our proposed control design resulted in over 80 %
that these buses can have almost equal impact on the reference of the improvement in the voltage profile of the MG compared to the base case
secondary control loop of the DG installed on bus 8. Moreover, from the scenario.
perspective of the load bus 4, both DGs installed on buses 2 and 3 have It can be seen that although the proposed heuristic design for the
almost the same topological position and therefore, the voltage devia­ control block cannot guarantee the minimum voltage deviation across
tion in bus 4 can take almost the same impact from the voltage the MG, it provided a significant improvement in the voltage profile
without dealing with the computational complexity of the optimization

9
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

problems. [4] Ramij Raja Hossain, Ratnesh Kumar, ‘A Distributed-MPC Framework for Voltage
Control Under Discrete Time-Wise Variable Generation/Load, IEEE Trans. Power
Syst. (2023).
5. Conclusion [5] N.M. Manousakis, G.N. Korres, Optimal PMU Placement for Numerical
Observability Considering Fixed Channel Capacity-A Semidefinite Programming
Approach, IEEE Trans. Power Syst. 31 (4) (Jul. 2016) 3328–3329, https://doi.org/
This paper proposed a novel approach to improve the voltage profile
10.1109/TPWRS.2015.2490599.
of islanded MGs using PMU devices and hierarchical control of distrib­ [6] M.A.R.S. Cruz, H.R.O. Rocha, M.H.M. Paiva, J.A.L. Silva, E. Camby, M.E.V. Segatto,
uted generators. We assumed PMUs are strategically placed across the PMU placement with multi-objective optimization considering resilient
microgrid based on existing methodologies to ensure observability of communication infrastructure, Int. J. Electr. Power Energy Syst. 141 (Oct. 2022)
108167, https://doi.org/10.1016/j.ijepes.2022.108167.
load buses. A three-level hierarchical control strategy was proposed [7] M.E. Hamedani Golshan, S.H. Hosseini Dolatabadi, S.M. Tabatabaei, Determining
utilizing PMU measurements at the tertiary control level to minimize minimum number and optimal placement of PMUs for fault observability in one-
voltage and frequency deviations. An "islanded tertiary control" struc­ terminal algorithms, IET Gener. Transm. Distrib. 12 (21) (Nov. 2018) 5789–5797,
https://doi.org/10.1049/iet-gtd.2018.5906.
ture was introduced to provide setpoints to the secondary control loops [8] S.H. Hosseini Dolatabadi, M.E. Hamedani Golshan, Fault location observability
of DGs to enhance the voltage profile of critical loads in islanded mode. rules for impedance-based fault location algorithms, Electr. Power Syst. Res. 224
Two design approaches including optimization-based and heuristic off­ (Nov. 2023) 109771, https://doi.org/10.1016/j.epsr.2023.109771.
[9] E. Khorram, M. Taleshian Jelodar, PMU placement considering various
line design were explained for the tertiary control and to reduce the arrangements of lines connections at complex buses, Int. J. Electr. Power Energy
computational complexity, a novel heuristic offline design was pro­ Syst. 94 (Jan. 2018) 97–103, https://doi.org/10.1016/j.ijepes.2017.06.028.
posed. Simulations on 30-bus, 39-bus, 57-bus, and 118-bus test networks [10] Yinghao Shan, et al., A unified model predictive voltage and current control for
microgrids with distributed fuzzy cooperative secondary control, IEEe Trans.
showed that the proposed hierarchical control strategy with "islanded
Industr. Inform. 17 (12) (2021) 8024–8034.
tertiary control" could improve the voltage profile of the islanded MGs [11] Dionysios Moutevelis, et al., Recursive Secondary Controller for Voltage Profile
by over 80 %. The heuristic offline design approach for the tertiary Improvement Based on Primary Virtual Admittance Control, IEEe Trans. Smart.
Grid. (2023).
control, based on simple rules considering network topology and DG
[12] Ahmed M. Abouassy, et al., Power Sharing Issues and Control Strategies in Islanded
locations, was able to achieve a significant improvement in voltage Microgrid: a Comparative Study, in: 2023 IEEE Conference on Power Electronics
profile leveraging available PMU measurements without complex opti­ and Renewable Energy (CPERE), IEEE, 2023.
mization and proved to be effective in significantly enhancing the [13] Hongxun Hui, et al., Coordination control of distributed generators and load
resources for frequency restoration in isolated urban microgrids, Appl. Energy 327
voltage profile of islanded MGs. The proposed approaches demonstrate (2022) 120116.
the immense potential of PMU-based control techniques to tackle key [14] Tianhui Meng, et al., Distributed frequency control of heterogeneous resources-
operational challenges in future MG systems. It is worth mentioning that powered islanded microgrid with enhanced economic operation, CSEE J. Power
and Energy Syst. (2022).
delays should be considered when applying the proposed hierarchical [15] Yang Mi, et al., A power sharing strategy for islanded DC microgrid with
structure, particularly with PMU-based signaling. Large signal delays unmatched line impedance and local load, Electric Power Syst. Res. 192 (2021)
could hamper fast inner control loops of converters. While the analysis 106983.
[16] Nohora España, et al., Distributed population dynamics for active and reactive
here focuses on foundational coordination, future work should incor­ power dispatch in islanded microgrids, Int. J. Electrical Power & Energy Syst. 125
porate communication and PMU delays into the models and control (2021) 106407.
design to ensure robustness. The presented studies represent an initial [17] Alghamdi Sultan, et al., Reduction in Voltage Harmonics of Parallel Inverters Based
on Robust Droop Controller in Islanded Microgrid, Mathematics 11 (1) (2023) 172.
analysis without delays, providing a basis for extensions accounting for
[18] Shuaicheng Hou, Jiawei Chen, Gang Chen, Distributed control strategy for voltage
real-world delays and latencies as future direction of this research. and frequency restoration and accurate reactive power-sharing for islanded
microgrid, Energy Reports 9 (2023) 742–751.
[19] J.C. Hernández, F. Sanchez-Sutil, P.G. Vidal, C. Rus-Casas, Primary frequency
CRediT authorship contribution statement
control and dynamic grid support for vehicle-to-grid in transmission systems, Int. J.
Electr. Power Energy Syst. 100 (Sep. 2018) 152–166, https://doi.org/10.1016/j.
Sayed Hamid Hosseini Dolatabadi: Writing – original draft, Visu­ ijepes.2018.02.019.
alization, Methodology, Conceptualization. Alireza Soleimani: Writing [20] P.G. Bueno, J.C. Hernández, F.J. Ruiz-Rodriguez, Stability assessment for
transmission systems with large utility-scale photovoltaic units, IET Renew. Power
– original draft, Methodology, Investigation, Conceptualization. Afshin Gener. 10 (5) (May 2016) 584–597, https://doi.org/10.1049/iet-rpg.2015.0331.
Ebtia: Writing – review & editing, Visualization, Validation, Method­ [21] F.J. Ruiz-Rodriguez, J.C. Hernández, F. Jurado, Voltage behaviour in radial
ology. Miadreza Shafie-khah: Writing – review & editing, Validation, distribution systems under the uncertainties of photovoltaic systems and electric
vehicle charging loads, Int. Trans. Electr. Energy Syst. 28 (2) (Feb. 2018) e2490,
Funding acquisition, Formal analysis. Tanveer Hossain Bhuiyan: https://doi.org/10.1002/etep.2490.
Supervision. [22] Fei Feng, Peng Zhang, Enhanced microgrid power flow incorporating hierarchical
control, IEEE Trans. on Power Syst. 35 (3) (2020) 2463–2466.
[23] Mohammad S. Golsorkhi, David J. Hill, Mehdi Baharizadeh, ‘A secondary control
Declaration of competing interest method for voltage unbalance compensation and accurate load sharing in
networked microgrids, IEEe Trans. Smart. Grid. 12 (4) (2021) 2822–2833.
The authors declare that they have no known competing financial [24] Jinshuo Su, et al., Membership-function-based Secondary Frequency Regulation for
Distributed Energy Resources in Islanded Microgrids with Communication Delay
interests or personal relationships that could have appeared to influence Compensation, IEEe Trans. Sustain. Energy (2023).
the work reported in this paper. [25] V. Mallemaci, F. Mandrile, S. Rubino, A. Mazza, E. Carpaneto, R. Bojoi,
A comprehensive comparison of Virtual Synchronous Generators with focus on
virtual inertia and frequency regulation, Electr. Power Syst. Res. 201 (Dec. 2021)
Data availability 107516, https://doi.org/10.1016/j.epsr.2021.107516.
[26] A. Suvorov, A. Askarov, A. Kievets, V. Rudnik, A comprehensive assessment of the
Data will be made available on request. state-of-the-art virtual synchronous generator models, Electr. Power Syst. Res. 209
(Aug. 2022) 108054, https://doi.org/10.1016/j.epsr.2022.108054.
[27] T.-L. Nguyen, Y. Wang, Q.-T. Tran, R. Caire, Y. Xu, C. Gavriluta, A Distributed
Hierarchical Control Framework in Islanded Microgrids and Its Agent-Based Design
References for Cyber–Physical Implementations, IEEE Trans. Ind. Electron. 68 (10) (Oct. 2021)
9685–9695, https://doi.org/10.1109/TIE.2020.3026267.
[28] Yang Han, Pan Shen, Xin Zhao, Josep M. Guerrero, Control strategies for islanded
[1] M. Heidari, et al., Exploring synergistic ecological and economic energy solutions
microgrid using enhanced hierarchical control structure with multiple current-loop
for low-urbanized areas through simulation-based analysis, AIMS Energy 12 (1)
damping schemes, IEEe Trans. Smart. Grid. 8 (3) (2015).
(2024) 119–151, https://doi.org/10.3934/energy.2024006.
[29] S. Choi, A.P.S. Meliopoulos, Effective Real-Time Operation and Protection Scheme
[2] S.Z. Paylakhi, N. Pariz, M.H.J.D. Bayaz, et al., Distributed secondary control of
of Microgrids Using Distributed Dynamic State Estimation, IEEE Trans. Power
islanded microgrids using modified integral backstepping technique, Int. J. Dynam.
Deliv. 32 (1) (Feb. 2017) 504–514, https://doi.org/10.1109/
Control 10 (2022) 2011–2019, https://doi.org/10.1007/s40435-022-00917-5.
TPWRD.2016.2580638.
[3] Sonam Shrivastava, Bidyadhar Subudhi, Comprehensive review on hierarchical
[30] Mohamed E. El-Hawary, Advances in Electric Power and Energy: Static State
control of cyber-physical microgrid system, IET Generation, Transmission &
Estimation, John Wiley & Sons, 2020.
Distribution 14 (26) (2020) 6397–6416.

10
S.H.H. Dolatabadi et al. Electric Power Systems Research 231 (2024) 110270

[31] Power Systems Test Case Archive, IEEE 14 Bus System, University of Washington, Syst. 32 (4) (Jul. 2017) 3258–3265, https://doi.org/10.1109/
2023 [Online]. Available: https://labs.ece.uw.edu/pstca/pf14/pg_tca14bus.htm TPWRS.2016.2616385.
[Accessed Feb 20,]. [34] Power Systems Test Case Archive, IEEE 57 Bus System, University of Washington,
[32] Power Systems Test Case Archive, IEEE 30 Bus System, University of Washington, 2023 [Online]. Available: https://labs.ece.uw.edu/pstca/pf57/pg_tca57bus.htm
2023 [Online]. Available: https://labs.ece.uw.edu/pstca/pf30/pg_tca30bus.htm [Accessed Feb 20,].
[Accessed Feb 20]. [35] Power Systems Test Case Archive, IEEE 118 Bus System, University of Washington,
[33] A.B. Birchfield, T. Xu, K.M. Gegner, K.S. Shetye, T.J. Overbye, Grid Structural 2023 [Online]. Available: https://labs.ece.uw.edu/pstca/pf118/pg_tca118bus.htm
Characteristics as Validation Criteria for Synthetic Networks, IEEE Trans. Power [Accessed Feb 20].

11