You are on page 1of 12

Energy Reports 7 (2021) 8030–8041

Contents lists available at ScienceDirect

Energy Reports
journal homepage: www.elsevier.com/locate/egyr

Coordinated intelligent frequency control incorporating battery


energy storage system, minimum variable contribution of demand
response, and variable load damping coefficient in isolated power
systems
∗ ∗∗
Amer Al-Hinai a,b , , Hajer Alyammahi c , Hassan Haes Alhelou a,b ,
a
Sustainable Energy Research Center, Sultan Qaboos University, Muscat, Oman
b
Electrical and Computer Engineering Department, Sultan Qaboos University, Muscat, Oman
c
Electrical and Computer Science Department, Khalifa University, Abu Dhabi, United Arab Emirates

article info a b s t r a c t

Article history: Maintaining a generation-demand balance becomes more challenging nowadays due to the limited
Received 16 May 2021 availability of traditional automatic generation control (AGC) and spinning reserves. In response
Received in revised form 12 July 2021 to this concern, the purpose of the proposed paper is to bridge the gap by introducing battery
Accepted 26 July 2021
energy storage system (BESS) control loop in the load frequency control (LFC) system. Furthermore,
Available online 9 August 2021
a minimum variable share of demand response (DR) is utilized as an additional mean to regulate the
Keywords: system frequency. The growing intermittency of controllable loads calls the awareness to incorporate
Load frequency control frequency-sensitive loads by considering a variable load-damping coefficient in LFC problem. The
Demand response presented control scheme demonstrates a stable power operation with the optimal sharing of BESS,
Frequency-sensitive load dynamic DR, and the supplementary control loop. Moreover, because of the continuous and sudden
Load-damping coefficient variations in the demand, adaptive LFC controller is designed using fuzzy logic based on particle swarm
Battery energy storage system optimization (PSO) tuning method. The proposed controller ensures a stable and robust frequency
Particle swarm optimization
regulation of the system under load fluctuations. In this paper it is revealed that adding the BESS
Fuzzy logic control
control loop is very effective in enhancing the performance of LFC as it can deliver fast power
compensation. The obtained results prove the capability and effectiveness of the proposed method
in an isolated power system.
© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction fluctuations in the demand-side due to the stochastic behav-


ior of consumers (Tungadio and Sun, 2019). This fact would
The balance services that keep the generation equal to the lead to additional costs and extra pollution, affiliated with the
demand in real time are of significant importance to operate required reserve from traditional power plants (Anon., 2006).
power systems stably, reliably, and securely. Traditionally, this Thus, Demand Response (DR) and storage devices are considered
balance is attained by controlling the amount of active-power to be alternatives to the conventional reserves to maintain the
generation, and ancillary services including both spinning and generation-load balance and sustain the system reliability in the
non-spinning reserves. With the increase the share of renewable case of sudden disturbances.
energy resources (RERs), more primary and secondary active- In conventional power systems, the frequency is controlled by
power reserves are required in the future in order to mitigate services called ancillary services usually provided by generation-
the impacts of the high fluctuations in generation and demand side. There are three level of controllers, i.e. primary, secondary,
sides. Unlike conventional power systems, future smart grids and tertiary control loops, where they are associated with pri-
would receive fluctuations from generation-side due to intermit- mary, secondary and tertiary reserves, respectively. It is clear that
tency in renewable energy sources in addition to the traditional consumers do not participate in providing frequency control and
regulation services. In fact, they can contribute to frequency pro-
∗ Corresponding author at: Electrical and Computer Engineering Department, tection schemes by under frequency load shedding (UFLS) service
Sultan Qaboos University, Muscat, Oman.
which is activated under contingency situations only. Considering
∗∗ Corresponding author. the fact that the demand-side is more flexible in comparison with
E-mail addresses: hinai@squ.edu.om (A. Al-Hinai), h.alhelou@squ.edu.om generation-side, therefore the responsive loads can participate in
(H. Haes Alhelou). real — time to provide ancillary services and to instantaneously

https://doi.org/10.1016/j.egyr.2021.07.072
2352-4847/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

respond to the area control error (ACE) signal (Mahto et al., Pan and Liaw, 1989; Vajk et al., 1985; Alquthami et al., 2020).
2021). Electric water heaters (EWHs) are considerable candidates Basically, the objective of having adaptive control in LFC is to
for the dynamic demand response as they can storage their hot increase the system robustness and make it less sensitive to the
water for a long time (Heffner et al., 2006; Hohne et al., 2019), sudden variations. For the above reason, a number of studies
which will presumably maintain the consumers’ quality of service have addressed the adaptive controller for modern power sys-
(QoS). In particular, all the smart appliances that storage their tems (Yousef et al., 2014; Srinivasa, 2012; Zhan et al., 2009;
energy thermally can be considered in demand side programs Al-Omari and Al-Hinai, 2007) as well as for islanded microgrids
including demand response services because their participation with penetrating renewable energy sources (Al-Omari and Al-
in demand response do not affect the comfort of consumers. For Hinai, 2007; Yufei et al., 2014). It is clear that the aforementioned
instance, appliances such as AC air-conditioners, water heaters, studies have suggested incorporating demand response or energy
and refrigerators are the most suitable electric smart devices that storages separately and have not suggested control schemes that
can be considered for future smart demand response programs. can be adopted in reality which can be considered as real research
However, the main challenge in the implementation of demand gaps that need to be considered in future researches.
response in real-word power systems is the need for a special This paper proposes a novel optimal coordination of load
infrastructure of information and communication technologies frequency control schemes. The proposed scheme utilizes both
(ICTs). BESS and DR in order to minimize the mechanical and thermal
Nowadays, the need for integrating energy storages is essential stresses on the conventional generation units, meaning that with
for modern power systems in order to deal with the negative the deployment of the proposed DR and BESS the frequency
effects of renewables on the safe operation of such systems. can be regulated smoothly. Moreover, it aims to stabilize the
With the increase of the renewable shares in modern power system frequency by controlling the load-damping coefficient in
systems, the magnitude of frequency fluctuations has been highly LFC problem. An adaptive proportional–integral controller is also
increased where the operator would need a faster and more designed to guarantee the desired performance under sudden
suitable reserves in order to maintain the frequency. A such fast disturbances which can help stabilizing the grid under critical
reserve cannot be delivered from conventional power plants due conditions. The proposed control method is applied to a large-
to the dynamic and mechanical limitations such as ramp-rate scale bulk power system. Simulation results prove the effective-
constrains (Jia et al., 2007). Therefore, the most suitable solution ness of the proposed control scheme. The main contributions of
is to harvest such a reserve from demand response and energy the proposed control scheme are highlighted as follows:
storage systems, however there is economic limitations regarding
• The suggestion and implantation of the minimum variable
the deployment of energy storage systems while the demand
participation of DR control scheme.
response remains as the best choice. In the case of a serious load
disturbance or a loss of a generation unit, the system perfor-
• The consideration of several and multiple step of load dis-
turbances in the designed scheme.
mance would be influenced, which could possibly lead to a critical
condition. Hence, incorporating demand response and partially
• The proposal and utilization of a variable load-damping
coefficient.
energy storage devices enhances the system performance and
stability by activating and releasing energy from demand side
• The evaluation of different share allocation of BESS contri-
bution in LFC.
at an appropriate time. The main backup energy storage devices
• The proposal of the optimal share allocation among DR,
are battery energy storage system (BESS), pumped hydro storage
BESS, and the supplementary control.
(PSH), and compressed air energy storage (CAES) (Aghajani and
Ghadimi, 2018). The paper is organized as follows. Section 2 describes and
During the last decade, conventional system frequency re- evaluates the concept of the developed SFR model while in-
sponse (SFR) models have been modified to integrate various corporating BESS and minimum share of DR with an optimal
types of emerging power generation units including distributed share allocation scheme. The controller design of load frequency
generating (DG) units-based dispatchable and non-dispatchable control scheme is explained in Section 3. The simulation results
renewable energy sources (Dong-Jing and Li, 2008; Ibraheem of different scenarios are presented and discussed in Section 4,
and Kothari, 2005), and storage devices such as flywheel energy while Section 5 concludes.
storage systems, and BESS (Li et al., 2005). In additions to de-
velopments in SFR models, the concept of the control has only 2. Proposed methodology and formulation
focused on the generation units and their AGC services, while
demand side management has not been included thoroughly in 2.1. LFC-DR-BESS model
the impactful studies. Among the few works, the references Liao
et al. (2015), Huang and Li (2013) and Omara and Bouffard (2009) The spinning reserves from conventional sources have sev-
have examined the effect of load-damping coefficient on the eral constraints and dynamic limitations including the minimum
frequency response in power systems (Liao et al., 2015; Huang up/down-time, ramp up/down rate limits, and other mechan-
and Li, 2013; Omara and Bouffard, 2009). However, the load- ical and dynamic generation constraints (Bevrani, 2009; Kun-
damping coefficient is expressed as a constant in most studies, dur, 1994). Hence, the system frequency response (SFR) model
where in fact, it is a variable factor when frequency-sensitive in (Bevrani, 2009) is modified to include the variable
loads are controlled. The most recent contribution in the LFC load-damping coefficient where frequency-sensitive loads can be
problem is considering DR control loop into the conventional considered as controlled loads as shown in Fig. 1. Furthermore,
LFC model which was reported in Mahto et al. (2021). However, the minimum variable participation of DR and BESS control loop
the proposed LFC-DR model does not ensure consumers’ QoS by are included in the secondary frequency control loop. Thus, the
utilizing a fixed amount from DR and it only considers purely main objective of the suggested model, called LFC-DR-BESS, is
resistive loads. In addition, it does not consider the effect of DR to minimize the amount of controlled loads besides minimizing
in the face of multiple load disturbances in which the adaptive the dependency on the supplementary control. The designed
controller is required. Considerable attention has been devoted DR control methodology starts with high variations in load that
for more than a quarter of a century to consider adaptive control drop with time. Accordingly, consumers’ QoS is enhanced and a
schemes in power systems (Kanniah et al., 1984; Ross, 1966; higher amount of consumer loads is obtainable for future control,
8031
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 1. Block-diagram of the SFR model proposed for LFC-DR-BESS control loops.

if required. By adding BESS control loop to the traditional SFR Tg , Tt , TBESS are the time constants of the governor, turbine, and
model, the system reliability would be increased since BESS can BESS, respectively. R is the droop coefficient of the governor and
deliver fast power compensation. Moreover, the dependency on Td is the DR communication delay. The participation coefficients
the supplementary control will decrease which means that the of DR, BESS, and the supplementary control are: α DR , α BESS , and
mechanical and thermal stresses on the generation units will be α Supp ., respectively. Later in the coming subsection, the optimal
highly reduced, and that result in increasing the lifetime of those share (participation) allocation scheme will be described in more
equipment. details.
Similar to spinning reserves, DR and BESS provide ancillary DR resources can be adjusted instantaneously by the received
services with a specified magnitude and power flow direction. command signal from the aggregator or operator considering a
When the frequency deviation is positive, a certain amount of prescheduling based on ancillary services market. Therefore, the
the responsive loads (that are OFF) will be turned ON, whereas only delay that should be considered when dealing with DR is
BESS will start charging, meaning that the demand side will the communication delay, often known as latency that affects the
absorb the surplus active power led to increasing the frequency system dynamic performance (Mahto et al., 2021). It is worth
above its nominal value. On the other hand, once the frequency mentioning that DR type that considered in this research work
deviation is negative, it is required to turn OFF a certain amount includes the direct control loads. In addition, the practical imple-
of the responsive loads while BESS will switch to discharging mentation of BESS is not always ideal. The following limitations
mode (Tungadio and Sun, 2019). Therefore, the power balance have to be taken into consideration (Teleke et al., 2010):
equation in frequency domain (including variable load-damping (1) State of Charge (SOC): The SOC is the available capacity of
coefficient, DR and BESS) can be expressed as (Mahto et al., 2021; the battery given as a percentage of its rated value.
Kundur, 1994): (2) Deep Discharge: Battery lifetime decreases by increasing the
Depth of Discharge (DOD). Most cell chemistries cannot sustain
∆PT (s)−∆PL (s)+∆PDR (s)+∆PBESS (s) = 2H ·s·∆f (s)+(D+∆D)∆f (s) the deep discharge and possibly they will damage in the case
(1) of fully discharged. Hence, in order to increase the lifetime of a
battery, the maximum discharge current of the battery should be
where, ∆PT (s) – ∆PL (s) is incremental power mismatch; ∆PDR limited.
(s) is the change in DR (∆PDR > 0 when ∆f > 0, and the reverse is (3) The Economic Cost of BESS.
true); ∆PBESS (s) is the change in contribution of BESS (∆PBESS > 0 In this paper, the impact of SOC and DOD is investigated in the
when the battery is in discharging mode, and the reverse is true); results section. It is also worth mentioning that the main focus of
H is system inertia constant; s is Laplace transform operator; this research work is on the frequency control in power systems
∆f (s) is frequency deviation; and (D + ∆D) is the change in as viewed from the control center. Therefore, the lifecycle of BESS
load-damping coefficient. is not a concern to the operator of the system as the power
The block-diagram shown in Fig. 1 presents a well-known system stability and security issue receives the higher importance
system frequency response (SFR) model which is suitable for de- in the control center. The lifecycle of BESS is considered to be
signing frequency controllers and testing them in power systems. main concern in the planning and operation of such storage
The demand in power system comprises several types of devices. devices.
For purely resistive loads, the change in the frequency does not
affect the consumed active power. In contrast, inductive loads, 2.2. Optimal share allocation of load damping, DR, BESS, and sup-
e.g. electric machines, are sensitive to changes in the system plementary control in frequency regulation
frequency. Therefore, the change in electrical power consists from
the frequency dependent and independent loads (∆PDR + D∆f ). It has been verified in Huang and Li (2013) that having a
The developed LFC-DR-BESS model is depicted in Fig. 1 where variable load-damping coefficient in SFR model would increase
8032
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 2. Frequency deviation under different load damping coefficient values.

Fig. 3. Frequency deviation under different DR contribution.

the accuracy of the model as it has a great impact on frequency regulation, different share values are applied in LFC-DR model
response. Based on various simulations and experiments, load- and compared to the conventional LFC. Fig. 3 indicates that for
damping coefficient has been varied under load disturbance of a higher DR contribution, a slight improvement in the dynamic
0.01 p.u. As shown in Fig. 2, with higher contribution of frequency performance is attained. When 10% of the required control is
sensitive loads in stabilizing the system frequency, better perfor- provided by the supplementary loop and 90% (fixed) from DR, a
mance is attained. Accordingly, the load damping coefficient is minor improvement in the overshoot is noticed.
varied as D = D + 0.9D in this study. Hence, this paper aims to minimize the share of DR based
Authors in Mahto et al. (2021) divided the required secondary on the disturbance level using a modified adaptive hill climbing
frequency control between demand response and the supplemen- (AHC) scheme (Pourmousavi and Nehrir, 2011; Weidong and
tary control loops according to their price at real-time electricity Dunford, 2004). A high share of DR will be assigned when the
market. They defined the share of supplementary control, α , to be disturbance occurs, and this share decreases with time as the
between 0 and 1. Accordingly, DR share is (1 −α ) since the sum of frequency deviation reduces. This scheme is adaptive since the
the participation factors should be unity. The suggested scheme DR share is a function of the frequency deviation that tends
considers a fixed amount of DR to be controlled regardless of the to minimize the deviation with the highest possible effort. The
disturbance level. To observe the contribution of DR in frequency system frequency deviation is the input variable to the controller.
8033
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 4. Frequency response when fThreshold = 0.005 Hz.

Fig. 5. Frequency response when M = 0.2.

At every time step, t, once the frequency deviation exceeds the optimal values. Different thresholds for the frequency deviation
desired limit (i.e., |(∆f )| > fThreshold ) where fThreshold is the desired were applied starting from fThreshold = 0.005 to fThreshold = 0.05
threshold value for the frequency deviation, the DR share is with step size of 0.001. The result when fThreshold = 0.005 Hz, is
calculated as follows: presented in Fig. 4. It is clear that when M is set to be 0.2, it gives
the most acceptable result compared to other values of M. This
αDR (t) = αDR (t − 1) + ∆f × M (2)
conclusion is also true under the other threshold values for the
where αDR (t − 1) is the previous value of DR share (at step t − 1) frequency deviation. In Fig. 5, it is evident that by comparing the
and the perturbation factor is M ∗ ∆f where M is a constant used frequency response when M is fixed at 0.2, the best performance
to bound the change in αDR . For higher M values, the dynamic attained is when fThreshold = 0.005. Therefore, those values are
performance may improve, but this will disturb the customers’ evidently considered in this paper.
QoS since more responsive loads will be controlled. Moreover, Since DR cannot provide a great enhancement in the frequency
when the value of M is high, it will presumably increase the regulation, BESS is proposed to further improve the system per-
oscillations in the frequency deviation. formance. In this paper, αBESS is optimized by PSO and its optimal
In this study, fThreshold and M are tuned by PSO and their opti- value is 0.9. In order to examine the contribution of BESS in
mal values are 0.005 Hz, and 0.2, respectively. Several simulation frequency regulation, different share values were tested in LFC-
experiments have been carried out under a load disturbance of BESS model and compared with the conventional LFC. As proved
0.01 p.u in order to evaluate the system response and confirm the in Fig. 6, for a higher change in BESS contribution, a notable
8034
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 6. Frequency deviation under a step load disturbance of 0.01 p.u.

Fig. 7. The flowchart of the proposed share allocation scheme in frequency regulation.

improvement in the dynamic performance is obtained. When 10% the supplementary control. The flowchart of the proposed share
of the required control is provided by the supplementary loop allocation scheme is presented in Fig. 7 where:
and 90% from BESS, a significant enhancement in the settling time
• αsupp. + αDR + αBESS = 1
and the overshoot is observed. However, as discussed earlier, the
contribution of BESS is limited because of the SOC and DOD.
• αBESS = 0.9
• 0 ≤ αDR ≤ 0.1
After performing the above comprehensive studies, there is an
abundant evidence to fix the BESS contribution to 0.9. As a result, The aim of the proposed control scheme is to minimize the
the remaining share is 0.1, which will be shared between DR and dependency on the supplementary control as well as the DR.
8035
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Compared to the speed governor, the proposed control strategy exist as can be seen from (8). Moreover, with the availability
performs faster because of the rapid response of the electrical of DR and BESS in the LFC, more degrees of freedom for the
loads and BESS. frequency regulation is provided. In the case when the supple-
mentary control cannot be accessed, the DR and BESS control
2.3. Steady-state error evaluation of the proposed model loops can guarantee a zero frequency deviation at steady state,
if their resources are sufficient to match the changes in the
It is well known that the fastest frequency control is the droop demand. The control effort required for controlling the frequency
control. However, in the case of a significant contingency, for is split among the control loops of supplementary, DR, and BESS
example the loss of a large generating unit, the droop character- as described previously.
istic is not enough to guarantee a zero steady state error in the
frequency deviation. As a result, the supplementary control loop 3. The design of the proposed control scheme
is essential to bring the system frequency back to its nominal
value (Kundur, 1994). The DR and BESS control loops are inte- Traditionally, the secondary frequency control, known as LFC
grated in this study for further control. Hence, it is required to or AGC, is usually implemented by a conventional PI controller.
evaluate the effect of the DR and BESS control loops on the steady These controllers are usually tuned based on trial and error ap-
state error. proach and in the best situation adjusted based on pre-knowledge
Eq. (1) can be rearranged as follows: and experiences of the operator. The main reason for incorpo-
1 rating PI controllers into the LFC is to effectively regulate the
∆f (s) = [∆PT (s)−∆PL (s)+∆PDR (s)+∆PBESS (s)], system frequency deviation to zero. The integral term is essential
2H · s + (D + ∆D)
to guarantee the zero steady state error while the proportional
(3) term is mainly needed to reduce the overshoot and enhance the
where ∆PT B(s) · [∆PSupp. (s) − 1
∆f (s) ], system dynamic performance (Goodwin et al., 2001). Many classi-
R
cal control methods have been implemented to tune PI controllers
1 like the well-known Ziegler–Nichols, Cohen–Coon, and Internal
B(s) = , (4)
(1 + s · Tg )(1 + s · Tt ) Model Control (Meshram and Kanojiya, 2012; Astrom and Hag-
glund, 2005). The main limitation of the previous methods is
By substituting (4) into (3):
their dependency on the system parameters. If the system operat-
1 1 ing conditions will change, the conventional controllers (e.g. PI)
∆f (s) = [B(s) · [∆PSupp. (s) − ∆f (s)]
2H · s + (D + ∆D) R (5) will be unable to provide sufficient control services in order to
− ∆PL + ∆PDR (s) + ∆PBESS (s)] achieve suitable dynamic performance. On the other hand, we
propose and adaptive control approach which can enhance the
By rearranging (5): performance of the proposed adaptive controller in this paper
B(s) continuously track the fluctuations in the power system and re-
∆f (s)[2H · s + (D + ∆D) + ] = B(s) · ∆PSupp. (s) − ∆PL (s) ject them. Such a controller will lead to the optimum performance
R (6)
+ ∆PDR (s) + ∆PBESS (s) required by the power system operator. Fuzzy logic has been
utilized as an intelligent tool for online adjusting of the controller
Solving (6) for ∆f (s): for over twenty years with achieving good control results (Zadeh,
1 1965).
∆f (s) = In this section, a PI-based intelligent controller is suggested
2H · s + (D + ∆D) +
B(s)
R
and designed using the combination of PSO and fuzzy logic. The
· [B(s) · ∆PSupp. (s) + ∆PDR (s) + ∆PBESS (s)] (7) fuzzy logic has the ability to compensate the shortages in the
1 traditional PI control theory especially for the complex systems
− · ∆PL (s)
2H · s + (D + ∆D) +
B(s) such as power systems where the level of uncertainties is high.
R
Generally, a fuzzy system has four parts, namely: fuzzy rule base,
∆P
where ∆PL (s) = S L ; the load disturbance is expressed as a unit fuzzification, inference system, and defuzzification (Mendel and
step function. Mouzouris, 1997).
The steady state value of frequency deviation is attained as To implement fuzzy systems to adjust the controllers’ gains,
follows: a set of fuzzy rules and relations is needed where 18 fuzzy rules
[∆PSupp,ss + ∆PDR,ss + ∆PBESS ,ss − ∆PL,ss ] are suggested in this paper to map the load fluctuation (∆P L ) and
∆fss = lim s · ∆f (s) = deviation of the frequency (∆f ) as input variables to the fuzzy
s→0 [2H · s + (D + ∆D) + B(s)
R
] system. Likewise, the outputs of the fuzzy system are defined
(8) as proportional gain (Kp ) and integral gain (Ki ), respectively. The
fuzzy rules adopted in this paper are provided in Table 1. The
where proposed fuzzy PI controller has the ability to improve the dy-
∆PSupp,ss = lim s · B(s) · ∆PSupp. (s) (9) namic performance of the system, where this promising result is
s→0 true based on a comparison with the other controllers. It is worth
∆PDR,ss = lim s · ∆PDR (s) (10) mentioning that the performance of fuzzy controllers is strongly
s→0
relied on the well-design of the membership functions. It can be
∆PBESS ,ss = lim s · ∆PBESS (s) (11) stated that the optimal selection of the relationships and fuzzy
s→0
rules has the most of the impact on the controller performance.
B(s) B(0)
[2H · s + (D + ∆D) + ]S =0 = (D + ∆D) + Therefore, an algorithm is required to optimally select and tuning
R R the membership functions which can guarantee the best dynamic
1 performance. This is necessary especially for systems with wide
≈ (D + ∆D) + (12)
R range of operating conditions, like power systems. In literature,
The steady state value for the frequency deviation will be zero PSO has been used successfully in power system controllers de-
if, and only if, the supplementary and/or DR and/or BESS controls sign since it guarantees a high accuracy result. It is also easy to
8036
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 8. The adaptive PSO based fuzzy PI controller in the closed-loop system.

Fig. 9. (a) Variable load disturbance. (b) Frequency response under variable load disturbance.

implement and it requires less optimization time compared to Results showed that the adaptive PSO-fuzzy PI controller had
other optimization methods. The optimal response is achieved a significant improvement in the microgrid system frequency
from any given initial guess. Therefore, in this paper the PSO is compared to the conventional PI controller (tuned by Ziegler–
utilized for adjusting the fuzzy PI controller membership func- Nichols) and the fuzzy PI controller in the presence of multiple
tions. The PSO algorithm has been described in Zhan et al. (2009),
load disturbances. Additionally, the conventional controller as
Al-Omari and Al-Hinai (2007), Al-Hinai and Al-Hinai (2009). The
adaptive PSO based fuzzy PI control structure (adaptive LFC) is well as the fuzzy PI controller could not manage the integration
shown in Fig. 8. of RESs due to their intermittency. In the contrast, the proposed
The robustness of the adaptive PSO based fuzzy PI controller adaptive controller guaranteed a stable frequency response under
is proved in an earlier work (Al Yammahi and Al-Hinai, 2015). the variable wind and photovoltaic generation.
8037
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 10. Frequency deviation for adaptive LFC-DR-BESS.

It is worth mentioning that the main focus of this research is secondary frequency control, a superior performance has been
on the frequency control in modern power systems, therefore, the achieved. A significant improvement in the overshoot as well as
details related to adaptive PSO is omitted since it is not the main the settling time is attained. Additionally, in Fig. 11, it is proved
contribution to the research. A comprehensive details related to that the objective of the proposed control scheme is met where
PSO can be found in Zhan et al. (2009), Al-Omari and Al-Hinai the contribution from the supplementary control in stabilizing
(2007), and Al-Hinai and Al-Hinai (2009). The required data for the microgrid frequency is significantly reduced.
tuning the PSO is given in Appendix.
The next section presents the contribution of minimum DR 4.3. Scenario III: Limitation of BESS
and BESS utilizing the adaptive fuzzy PI controller tuned based
an evolutionary algorithm, i.e. PSO. In the first scenario, when the load increases suddenly by
3% at t = 8 s as demonstrated in Fig. 9(a), the BESS becomes
4. Simulation results and discussion fully discharged at t = 10 s. The system frequency response is
displayed in Fig. 12. By comparing the adaptive LFC-DR-BESS with
Simulation experiments have been carried out to the islanded limited and unlimited BESS SOC, it is clear that the limitation of
microgrid, shown in Fig. 1, to evaluate the impacts of the mini- the BESS capacity affects the system frequency response. Even
mum DR share and BESS contribution in the secondary frequency though the BESS becomes fully-discharged, the proposed adaptive
control as well as to illuminate the effectiveness of the proposed LFC-DR-BESS guarantees a better performance compared to the
coordinated LFC scheme. The rated power of the generation unit adaptive LFC model where the minimum share of DR compen-
(supplementary control) and BESS are listed in Table 2 in Ap- sates the loss of BESS. On the other hand, when the load abruptly
pendix. The studied microgrid system parameters are listed in drops at t = 30 s by 5%, the BESS becomes fully-charged and
Table 3. the limits on its charging and discharging capacity is applied as
shown in Fig. 13. The deviation in system frequency response
4.1. Scenario I: LFC-min(DR) is illustrated in Fig. 12. By comparing the adaptive LFC-DR-BESS
model with limited and unlimited BESS contribution, it is notable
In this case, the adaptive LFC is compared to the proposed that with the limited capacity of BESS, DR plays an important role
adaptive LFC-min(DR) while considering a variable load distur- in enhancing the system frequency.
bance that is shown in Fig. 9(a). The system frequency response
is presented in Fig. 9(b). It is obvious that with the minimum DR 4.4. Discussion and future work
participation in frequency regulation, the dynamic performance
is enhanced in terms of the settling-time, overshoot, as well as in There is a great tendency for future power plants to integrate
minimizing the oscillations. Furthermore, the adaptive fuzzy PI numerous Distributed Generations (DGs), thus the problem that
controller tuned based on PSO reveals robustness of the system will arise is what will happen to the LFC requirements since the
against the dynamic and sudden disturbances. conventional controllers may not be able to withstand the contin-
uous variations and uncertainties in power generation. Moreover,
4.2. Scenario II: LFC-DR-BESS in islanded microgrids, the system challenges increase and the
need for having a complementary control loop is vital.
In this scenario, the same system is studied after incorporating The current work can be considered as a solution to the
the BESS control loop into the developed adaptive LFC-min(DR). It above issues and concerns where the adaptive control scheme
can be seen in Fig. 10 that due to the contribution of BESS in the has been developed with integrating DR and BESS control loops
8038
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 11. Contribution from the supplementary control loop.

Fig. 12. BESS limitation in adaptive LFC-DR-BESS.

into the LFC. In this paper, we have mainly focused on regulat- 5. Conclusion
ing the system frequency while minimizing the dependency on
the conventional spinning reserves. DR and BESS show a great Conventional controllers are commonly unable to provide a
enhancement in balancing generation and demand. In addition, good performance in the presence of serious disturbances. In
response to this issue, an adaptive PI controller is utilized for LFC
they help in avoiding the use of the conventional AGC which is
problem in the present paper using fuzzy logic based PSO tuning
environmentally unfriendly. method. Adaptive fuzzy controller has a good performance com-
It is proven that the adaptive PSO based fuzzy controller pared to the conventional control methods. However, the optimal
performs an important role in managing the short-term dis- performance is not assured. In order to guarantee the optimal
turbances. Moreover, with the minimum DR participation, the performance under a wide operating range, the PSO algorithm is
system performance enhances and consumers’ QoS is maintained. used to optimally choose the membership functions. This paper
BESS also added a great improvement in frequency regulation. proposes a further control scheme by introducing the minimum
However, the contribution of BESS is limited due to the SOC and participation of DR and BESS into the LFC problem. The developed
control scheme has been implemented for a single area power
DOD limitations as discussed earlier. So, with the availability of
system. The balance between generation and demand is achieved
the intelligent controller as well as DR in the LFC, more degree by utilizing BESS and a percentage of available responsive loads
of freedom for the frequency regulation is achieved. Without and the supplementary control, based on the suggested optimal
having the proposed control scheme, it may be very challenging share allocation scheme. The main objective of the proposed
to integrate large number of DGs into the power system. control strategy is to replace the costly conventional reserves
8039
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Fig. 13. Limitation of BESS contribution.

with DR and BESS. It is well known that storage devices as well as Table 2
DR can compensate the variations and the uncertainties in power Rated power of generation unit and BESS units and loads.

system. Additionally, they show a great promise in enhancing Rated Power (KW) Load (KW)
the system efficiency. It is revealed through different simulation Generation unit 160 PL
180
studies that the proposed control scheme improves the overall BESS 45

performance of the system frequency.


Table 3
CRediT authorship contribution statement Power system parameters (Zhan et al., 2009; Bevrani, 2009).
Tg Tt R 2H D Td TBESS
Amer Al-Hinai: Conceptualization, Data curation, Methodol- 0.08 s 0.4 s 3.0 Hz/p.u. 0.1667 p.u. s 0.015 p.u./Hz 0.1 s 0.1 s
ogy, Formal analysis, Writing – original draft, Writing – review
& editing. Hajer Alyammahi: Conceptualization, Data curation,
Methodology, Simulation experiments Formal analysis, Writing –
original draft. Hassan Haes Alhelou: Data curation, Methodology, In what follows, the used data in tuning PSO is given: Popu-
Formal analysis, Writing – review & editing, Funding, Project lation Size = 11; Numbers of iterations = 50; Inertia Weight (w)
administration. = 0.8; Cognitive Coefficient (C1) = 2; Social Coefficient (C2) = 2.

Declaration of competing interest References

The authors declare that they have no known competing finan- Aghajani, G., Ghadimi, N., 2018. Multi-objective energy management in a
cial interests or personal relationships that could have appeared micro-grid. Energy Rep. 1 (4), 218–225.
to influence the work reported in this paper. Al-Hinai, A.S., Al-Hinai, S.M., 2009. Dynamic stability enhancement using par-
ticle swarm optimization power system stabilizer. In: 2nd International
Conference on Adaptive Science & Technology 2009. ICAST 2009, pp.
Acknowledgment 117–119.
Al-Omari, B.S., Al-Hinai, A.S., 2007. Design of fuzzy-swarm load frequency
This work was supported by his majesty (HM) Fund and Sultan controller for interconnected power system. In: 4th IEEE GCC Conference
Qaboos University (SQU) under Grant SR/ENG/ECED/17/1. and Exhibition. Manama, Bahrain.
Al Yammahi, H., Al-Hinai, A., 2015. Intelligent frequency control using optimal
tuning and demand response in an AC microgrid. In: 2015 International
Appendix
Conference on Solar Energy and Building (ICSoEB). pp. 1–5. http://dx.doi.
org/10.1109/ICSoEB.2015.7244943.
The data and information that used for setting up the simula- Alquthami, T., Butt, S.E., Tahir, M.F., Mehmood, K., 2020. Short-term opti-
tion. Table 1 presents the fuzzy relationships. The rated powers mal scheduling of hydro-thermal power plants using artificial bee colony
of generation unit, BESS, and loads are given in Table 2. Ta- algorithm. Energy Rep. 1 (6), 984–992.
ble 3 presents the important data for the isolated system under Anon., 2006. Benefits of demand response in electricity markets and recom-
mendations for achieving them: A report to U.S. Congress. U.S. Dept. Energy,
investigation.
[Online]. Available: http://eetd.lbl.gov/ea/ems/reports/congress-1252d.pdf.
Astrom, K.J., Hagglund, T., 2005. Advanced PID Control. ISA - The Instrumenta-
tion, Systems, and Automation Society.
Table 1 Bevrani, H., 2009. Robust Power System Frequency Control. Springer, NewYork,
The fuzzy rules set. NY, USA (ch. 1–3).
∆P /∆f NL NM NS PS PM PL Dong-Jing, L., Li, W., 2008. Small-signal stability analysis of an autonomous
hybrid renewable energy power generation/energy storage system Part I:
S NL NM NS PS PS PM
Time-domain simulations. IEEE Trans. Energy Convers. 23 (1), 311–320.
M NL NL NM PS PM PM
Goodwin, G.C., Graebe, S.F., et al., 2001. Control System Design. Prentice Hall,
L NL NL NL PM PM PM
Upper Saddle River, N.J..

8040
A. Al-Hinai, H. Alyammahi and H. Haes Alhelou Energy Reports 7 (2021) 8030–8041

Heffner, G.C., Goldman, C.A., Moezzi, M.M., 2006. Innovative approaches to Omara, H., Bouffard, F., 2009. A methodology to study the impact of an
verifying demand response of water heater load control. IEEE Trans. Power increasingly nonconventional load mix on primary frequency control. In:
Deliv. 21 (1), 388–397. Proc. IEEE PES General Meeting.
Hohne, P.A., Kusakana, K., Numbi, B.P., 2019. A review of water heating tech- Pan, C.T., Liaw, C.M., 1989. An adaptive controller for power system and load
nologies: An application to the South African context. Energy Rep. 1 (5), frequency control. IEEE Trans. Power Syst. 4 (1), 122–128.
1–9. Pourmousavi, S.A., Nehrir, M.H., 2011. Demand response for smart microgrid:
Huang, H., Li, F., 2013. Sensitivity analysis of load-damping characteristic Initial results. In: Proc. 2nd IEEE PES Innov. Smart Grid Technol.. ISGT, pp.
in power system frequency regulation. IEEE Trans. Power Syst. 28 (2), 1–6.
1324–1335. Ross, C.W., 1966. Error adaptive control computer for interconnected power
Ibraheem, P.K., Kothari, D.P., 2005. Recent philosophies of automatic generation system. IEEE Trans. Power Appar. Syst. PAS-85, 749.
control strategies in power systems. IEEE Trans. Power Syst. 20 (1), 346–357. Srinivasa, C., 2012. Adaptive neuro fuzzy based load frequency control of multi
Jia, R., Nehrir, M.H., Pierre, D.A., 2007. Voltage control of aggregate electric water area system under open market scenario. In: International Conference on
heater load for distribution system peak load shaving using field data. In: Advances in Engineering Science and Management. ICAESM, pp. 5–10.
Proc. 39th North Amer. Power Symp.. NAPS, pp. 492–497. Teleke, S., Baran, M.E., Bhattacharya, S., Huang, A.Q., 2010. Optimal control
Kanniah, J., Tripathy, S.C., Malik, O.P., 1984. Microprocessor based adaptive load of battery energy storage for wind farm dispatching. IEEE Trans. Energy
frequency control. Proc. Inst. Electr. Eng. C 131 (4), 121–128. Convers. 25 (3), 787–794.
Kundur, P., 1994. Power system Stability and Control. McGraw-Hill, NewYork, Tungadio, D.H., Sun, Y., 2019. Load frequency controllers considering renewable
NY, USA (ch. 11).
energy integration in power system. Energy Rep. 1 (5), 436–453.
Li, Y.H., Choi, S.S., Rajakaruna, S., 2005. An analysis of the control and operation
Vajk, I., Vajta, M., Keviczky, L., 1985. Adaptive load frequency control of
of a solid oxide fuel-cell power plant in an isolated system. IEEE Trans.
Hungarian power system. Automatica 21 (2), 129–137.
Energy Convers. 20 (2), 381–387.
Weidong, X., Dunford, W.G., 2004. A modified adaptive hill climbing MPPT
Liao, S., et al., 2015. Load-damping characteristic control method in an isolated
method for photovoltaic power systems. In: Power Electronics Specialists
power system with industrial voltage-sensitive load. IEEE Trans. Power Syst.
Conference, vol. 3, pp. 1957–1963.
99, 1–11.
Yousef, H.A., Al-Kharusi, K., Albadi, M.H., Hosseinzadeh, N., 2014. Load frequency
Mahto, T., Malik, H., Mukherjee, V., Alotaibi, M.A., Almutairi, A., 2021. Renewable
control of a multi-area power system: An adaptive fuzzy logic approach.
generation based hybrid power system control using fractional order-fuzzy
Power systems. IEEE Trans. Power Syst. 29 (4), 1822–1830.
controller. Energy Rep. 1 (7), 641–653.
Yufei, T., Tang, J., Yang, J., Haibo, H., Yan, Z., 2014. Frequency control using on-
Mendel, J.M., Mouzouris, G.C., 1997. Designing fuzzy logic systems. IEEE Trans.
Circuits Syst. II 44, 885–895. line learning method for island smart grid with EVs and PVs. In: International
Meshram, P.M., Kanojiya, R., 2012. Tuning of PID controller using Ziegler-Nichols Joint Conference on Neural Networks. pp. 1440–1446.
method for speed control of DC motor. In: Advances in Engineering, Science Zadeh, A., 1965. Fyzzy sets. Inf. Control 8, 338–353.
and Management, 2012 International Conference On. ICAESM, pp. 117–122. Zhan, Z.H., Zhang, J., Li, Y., Chung, H.S.H., 2009. Adaptive particle swarm
optimization. IEEE Trans. Syst. Man Cybern. B 39 (6), 1362–1381.

8041

You might also like