Available online at www.sciencedirect.

com
ScienceDirect

Energy Reports 6 (2020) 1597–1603

www.elsevier.com/locate/egyr

2020 7th International Conference on Power and Energy Systems Engineering (CPESE 2020),
26–29 September 2020, Fukuoka, Japan

Load frequency control for renewable energy sources for isolated

power system by introducing large scale PV and storage battery
Lei Liua ,∗, Tomonobu Senjyua , Takeyoshi Katob , Abdul Motin Howladerc , Paras Mandald ,
Mohammed Elsayed Lotfye,a
a
Department of Electrical and Electronics Engineering, University of the Ryukyus, Okinawa 903-0123, Japan
b Institute of Materials and Systems for Sustainability, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Japan
c Smart Grid Research Lab, University of California, Riverside, CA, 92521, USA
d Department of Electrical and Computer Engineering, The University of Texas at EI Paso, Texas, TX 79968, USA
e Department of Electrical Power and Machines, Zagazig University, Zagazig 44519, Egypt

Received 13 December 2020; accepted 14 December 2020

Abstract
In this work we had presented a power system electricity frequency control approach of 100% renewable energy sources
isolated power grid by applying model predictive control (MPC). As the PV panel and storage battery price had dramatically
dropped in these years, large capacity of PV generation and storage battery had been introduced base on an original 100%
renewable energy power system. MPCs had been design for each of the model (Photovoltaic Generation, Wind turbine
Generation, Storage Battery, Fuel Cell and Seawater Electrolyzer) to suppress the consumption and generation balance. In
addition, in order to suppress the load frequency, demand response (real-time pricing) also had been applied in this scheme
by adjust the controllable loads. All of the control approach proposed in this work had been verified by Matlab@/Simulink@.
And by discuss and analysis the simulation results, the stability, robustness, reliability and effectiveness of this scheme had
been examined.
⃝c 2020 The Author(s). 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/).
Peer-review under responsibility of the scientific committee of the 7th International Conference on Power and Energy Systems Engineering, CPESE,
2020.

Keywords: Model predictive control; Renewable energy sources; Photovoltaic Generation; Storage battery; Supply–demand balance; Real-time
pricing; Load frequency control

1. Introduction
Recent decades, within the renewable energy sources (RES) are developing all over the world, it had played
an increasingly important role of environmental issues especially greenhouse effect issue. And in the foreseeable
future decades, the renewable energy generation will get more attention and massive growing [1]. However, we also
∗ Corresponding author.
E-mail address: [email protected] (L. Liu).

https://doi.org/10.1016/j.egyr.2020.12.030
2352-4847/⃝ c 2020 The Author(s). 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/).
Peer-review under responsibility of the scientific committee of the 7th International Conference on Power and Energy Systems Engineering,
CPESE, 2020.
L. Liu, T. Senjyu, T. Kato et al. Energy Reports 6 (2020) 1597–1603

have to face lots of problems brought about by the developing of renewable energy generation. One of the RES
characteristics is it will change with objective natural conditions, particularly evident in Wind Generation (WG)
and Photovoltaic Generation (PV), etc. In the RESs power system, the natural condition like weather condition
will cause rapid deviation of RESs generation, which is the main reason for system supply and demand imbalance
and frequency changing. Therefore, many scholars had pay attention on find the control approaches to achieve the
balance between demand and response side of micro renewable energy power system [2,3].
One of the solutions is by use battery equipment to solve the load frequency deviation issue. However, the
battery solutions have many limitations such as battery capacity limit, unable to achieve long-term adjustment.
Furthermore, even considering demand-response in this control scheme, it is still difficult to solve RESs generating
unstable caused by frequent changes in weather conditions. Another solution is by using controllers to achieve load
frequency stable and minimization of the RESs generation loss. In this work, we applied MPC because it could
achieve the above functions through predictive in advance and emit control commands [4–6].
Based on the 100% RESs micro system we had proposed previous paper, at the same time by consider the sharp
drop in the price of solar panels and battery devices in recent years. In this paper, we attempt to Significantly
increase the PV generation capacity and storage battery capacity. Then by compare and analysis the simulation
results, to examine the stability, robustness, reliability and effectiveness of proposed control scheme.

2. System configuration
Fig. 1 shows the 100% RESs remote island grid, it include Wind Generation (15MW), Photovoltaic Generation
(20×2 MW), storage battery (10 MW/40 MWh×2), seawater electrolyzer (5 MW), fuel cell (5MW) and the load
assumed as an Energy Management System (EMS). And Table 1 is the proposed system parameters. In this power
grid model, MPC had been designed for each generation.

Table 1. System Parameters.

Parameters D M Twg Tpv Tb Tfc Tae
Contents 0.015 pu/Hz 0.1667 pu/s 1.5 s 1.8 s 0.1 s 4.0 s 0.5 s

3. Model predictive control

MPC is the advanced control strategy which could predict in advance and satisfies multiple constraint equations
to achieve the control target. Each prediction model had been designed according to the state–space equation show
in Eqs. (1) and (2). For all the controllers, sample time was set as 0.1 s, prediction horizon was set as 10 s, and
control horizon was set as 2s.
ẋ = Ax(t) + Bu(t) (1)

ẏ = C x(t) + Du(t) (2)

In the above state–space equation, A, B, C and D are real constant matrices to describe the proposed power
system within the appropriate dimensions. And u, x and y indicate the input vector, state vector, and the output
vector.

4. Control method
4.1. Real-time pricing

Depending on the different implementation scenarios, difficult types of electricity pricing methods have their
own strengths. Common methods of demand-response have Time-Of-Use (TOU) pricing and real-time pricing,
TOU could achieve peak shaving and valley filling by changing the electricity price within the specific time period.
The real-time pricing which applied in the work could by changing electricity prices in real time to achieve power
electricity supply–demand balance.
1598
L. Liu, T. Senjyu, T. Kato et al. Energy Reports 6 (2020) 1597–1603

Fig. 1. Linearized model of power system.

The price equation is shown below Eq. (3). The electricity pricing was calculated by Eq. (3) with frequency
deviation.
−0.3
π∗ = + 0.35 [Yen] (3)
1 + exp(−40 · △ f )
In this equation, the electricity was calculated by consider the demand-response balance and the frequency
deviation. Since the load frequency increases or the load consumption drops, the electricity price will decrease in
time to encourage the load consumption increase, and vice versa. And to adjust the rapid system power fluctuation,
a first order lag filter with time constant Tπ was applied to slow down the dramatic changes of electricity price.
Fig. 2 shows the change curve of electricity price with system frequency deviation ∆ f .

Fig. 2. Electricity price π for frequency deviation ∆ f .

1599
L. Liu, T. Senjyu, T. Kato et al. Energy Reports 6 (2020) 1597–1603

4.2. Auto demand response

The power consumption predetermined by electric price was shown in Fig. 3. Based on this curve, the consume
load corresponding electricity price automatically and achieved the energy management. In this paper, eight different
kinds of loads were roughly divided as three categories which had displayed in Fig. 3:

Fig. 3. Power consumption of each load for electric price.

(1) Loads 1, 2, 3, 4: The active power of these load continuous changes with following the price changing in
different rate;
(2) Loads 5, 6: In order to response the rapid price fluctuations, this part the consume loads proposed as hysteresis
characteristics in power consumption fluctuations. For example as load 5, since the electricity price increase until
30 yen, the active power of load drops to 0 pu; instead, the active power will not have any action if the power
electricity only decrease to 30 yen, only if it reach to 25 yen, the active power will increases to 0.05 pu.
(3) Load 7, 8: a different size of step hysteresis characteristic with comply with fluctuation of electricity
consumption.
Simulation Results
In this work, the simulation condition in this work was performed assuming that the solar radiation decreased
at t = 600–700 s.
And in this section, we discuss about the simulation results clarify the effectiveness and robust performance
of the work. First, Fig. 4 shows the frequency fluctuation, here we could found this control approach has good
performance, even it still have some disturbs caused by solar radiation, but it shows effectiveness and robustness of
a power system frequency control scheme of the renewable power model; Fig. 5 shows the generated and consumed
power, we can found as applied MPC in this work, the generated power was followed the consumed power, it mean
it keeps steady balance between supply side and demand side; Fig. 6 is the output power of battery; Fig. 7 shows
the consumed power of each load; Figs. 8 and 9 show the output power of PV generator and wind generator, from
these two results we can find because of the large capacity of PV, in this case the PV and wind output did not
followed the MPPT, they all work on a lower level state, but if we increase the load or complete the load model,
the PV generation and wind generation will follow the MPPT and help to compensate power balance; Figs. 10 and
11 are output power of fuel cell generation and seawater electrolyze generation, respectively.

Fig. 4. frequency deviation ∆ f .

1600
L. Liu, T. Senjyu, T. Kato et al. Energy Reports 6 (2020) 1597–1603

Fig. 5. Generated and consumed power, Pg, PL.

Fig. 6. Output power of battery, Pb .

Fig. 7. Consumed power of each DR load, PLoad .

Fig. 8. Generated power of photovoltaic generator, Ppv .

5. Conclusion

The paper shows a load frequency control scheme by apply MPC controllers in the power grid within large-
scale introduction of PV generation and storage battery, and the renewable energy generation facilities rate of
1601
L. Liu, T. Senjyu, T. Kato et al. Energy Reports 6 (2020) 1597–1603

Fig. 9. Generated power of wind turbine generator, Pwg .

Fig. 10. Generated power of fuel cell, Pfc .

Fig. 11. Consumed power of seawater-electrolyzer, Pae .

system model is 100%. The MPC was applied in this work to control the output generation of each model, include
storage batteries, PV, WG, fuel cell, and seawater electrolyzer. The controllers applied in each generation model
was determines the optimal control instruction by forecast the future generation output value based on the previous
data and information of each generation model. Particularly, in this research we had increased the capacity of PV
and storage battery to discuss the impact of large-capacity PV generation and storage battery on system frequency.
Via simulation results we confirmed the system frequency control goal was achieved by proposed control scheme.
In future work, we have to plan to improve this work by complete the load model, and also considering to using
a centralized MPC control scheme, it may solve the frequency disturb in this model, and also have much good
feedback performance in 100% renewable power model.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could
have appeared to influence the work reported in this paper.
1602
L. Liu, T. Senjyu, T. Kato et al. Energy Reports 6 (2020) 1597–1603

References
[1] Liu L, Matayoshi H, Lotfy ME, Datta M, Senjyu T. Load frequency control using demand response and storage battery by considering
renewable energy sources. Energies 2018;1(12):3412.
[2] Siano P. Demand response and smart grid - a survey. Renew Sustain Energy Rev 2014;30:461–78.
[3] Uebel S, Murgovski N, Bäker B, Sjöberg J. A two-level MPC for energy management including velocity control of hybrid electric
vehicles. IEEE Trans Veh Technol 2019;68(6):5494–505. http://dx.doi.org/10.1109/TVT.2019.2910728.
[4] Shiroei M, Ranjibar A. Supervisory predictive control of power system load frequency control. Int J Electr Power Energy Syst
2014;61:7080.
[5] Zhang Y, Liu X, Yan Y. Model predictive control for load frequency control with wind turbines. J. Control Sci. Eng. 2015;2015:17,
282740.
[6] Kayalvizhi S, Vinod Kumar DM. Load frequency control of an isolated micro grid using fuzzy adaptive model predictive control. IEEE
Access 2017;5(Aug.):16241–51.

1603