Sie sind auf Seite 1von 12

Classical controller based AGC of a hybrid

multisource power system incorporating

distributed generation
Cite as: AIP Conference Proceedings 2091, 020002 (2019);
Published Online: 02 April 2019

Debdeep Saha, Lalit Chandra Saikia, and Bipul Kumar Talukdar

AIP Conference Proceedings 2091, 020002 (2019); 2091, 020002

© 2019 Author(s).
Classical Controller Based AGC of a Hybrid Multisource
Power System Incorporating Distributed Generation
Debdeep Saha1,*, Lalit Chandra Saikia2 and Bipul Kumar Talukdar1
Department of Electrical Engineering, Girijananda Choudhury Institute of Management and Technology,
Department of Electrical Engineering, National Institute of Technology Silchar

*Corresponding e-mail:

Abstract. The article exhibits automatic generation control of an interconnected realistic multi-source two area system. The
system comprises of two thermal units in Area 1 and one hydro and one thermal unit in Area 2 and, incorporating distributed
energy generation (DG) in Area 1. Non-linearity’s such as appropriate generation rate constraint of 3%/min for thermal units,
270%/min for rising and 360%/min for lowering action of hydro units are incorporated along with governor dead band of 0.036
Hz for both thermal and hydro unit. Classical controllers are employed as secondary controller to reduce the area control error
due to the intermittent non-conventional sources in distributed generation and thereby reducing the frequency and tie-power
deviations. Stochastic fractal search (SFS) algorithm is utilized to tune the classical controller gains and other parameters
simultaneously. Performance comparison of classical controllers such as proportional-integral-derivative (PID) and proportional-
integral-derivative with first order filter effect (PIDN) in absence of DG unit establishes the superiority of PIDN controller over
PID. Simulations are carried out in presence of DG unit in Area 1 by incorporating wind turbine generator system (WTS) and
dish sterling solar thermal (DSTS) plant that are tested with step input as well as random input. The solutions obtained from the
robust SFS based PIDN controller ensures that the frequency and tie-power deviations are within safe limits when random pattern
of WTS and DSTS plant are applied in the DG unit. Also, a part of the total generation shared by the DG unit helps in balancing
the real power output and hence the decrease in deviations occurs.

An interconnected network aims to regulate frequency and tie-line power deviation automatically when there are any
changes in load rather than only using a speed governing mechanism. Automatic generation control (AGC) enables
the effective execution of the above. Thus, objective of AGC is to maintain the scheduled system frequency and tie-
line power. Frequent changes in Load demands show visible changes constantly, and therefore it is becomes
important to regulate balance between the amount of power generated and demanded. On the occasion of unhealthy
conditions, when there is a difference between total power generation and load demand, frequency and tie-power
changes from their nominal values, which causes fluctuations in the system dynamics [1].
The environmental concerns in the age of deregulated market have shifted the centralized power generation
to a more decentralized generation mode [2]. Thus there is a need for the distributed energy resources (DERs) with
integration of renewable energy technologies like wind and solar, along with energy storage devices like flywheels,
batteries, etc., and combined heat and power generation technologies [2-3]. In recent years, Smart grids and micro
grids, an unexplored area to study, are becoming important topics for energy demand and constitutes a mode to well
integrate various sources of distributed generation (DG), especially non-conventional energy sources, and thus
reduce CO2 and NOX emissions. It is to be noted considered that increasing levels of penetration of DGs may cause
severe phase voltage deviation, resulting in an excess ground current and degrading power quality. An organized
study with an efficient solution approach is vital to keep a reliable and economic quality of energy, thereby
motivating to control the frequency and power deviation in an interconnected system whose analysis is need of the
hour [4-5].
One of the constituents of DG is FCs (Fuel cells) whose electricity production comes with the advantages
of high efficiency, least pollution, onsite installation, reusability of exhaust heat and water, and the ability to use a

Current Trends in Renewable and Alternate Energy

AIP Conf. Proc. 2091, 020002-1–020002-11;
Published by AIP Publishing. 978-0-7354-1821-9/$30.00

wide range of fuels [4]. The research and development of the Dish Stirling Solar Thermal Plant (DSTS) can be
traced back to the 1990s, and being cheap and efficient leads to a surprisingly high growth in the installed capacity
in recent years. However, DSTS technology has relatively less energy conversion efficiency, lower power density,
and higher cost compared to wind turbine system (WTS) [5].
Researchers have proven that distributed generation (DG) has a vital role to play in the future of grid
systems along with energy storage and demand side management. The excess load caused due to sudden change in
load demand in an interconnected system can be shared by Distributed Energy resources, however, tending the
system instability towards stability mode. Funcke and Bauknecht [6] presented a hierarchy based on the position of
generating units and equipped methods of system balancing for describing visions of centralized and decentralized
electricity systems. Pepermans et al. reviewed available definitions [7] and categorizations of DG. A DG is always
connected to the customer side through network infrastructure and metered connections along with the generating
sources [8].
Generation using Wind, Tidal, solar, PV, biomass, cogenerating powers are sustainable generation by
nature. Among all of the above, wind and solar can play a major role in providing quality and reliable power and
hence, assume great importance [9-10]. Also, Fuel cell (FC) adds up to the list in providing green and clean power to
the grid. Off - grid electricity or isolated generation is possible by utilizing single source system such as using solar
photovoltaic panels, WTS, DSTS unit, micro - hydel plants, or co-generator sets, or by combining two or more of
these electricity producing sources which we term as a hybrid system. The systems often include energy storage in
the form of batteries. A hybrid system can supply power AC or DC or both [9]. Component or system control or
both is used to regulate the overall system operation [9-10, 11-12]. Das at al [9-10] has carried out the small signal
analysis of a hybrid power system in isolated condition considering the DG unit as one of the sources of generation.
Also, Rahman et al [13-14] has dealt with automatic generation control of an interconnected system considering
DSTS as one of the units of power sources.
Although, several works have been reported with DG units individually but no such work is being reported
in an interconnected system where frequency and tie-power fluctuations are tested. No work till date has considered
DG unit with thermal and hydro unit where the impact of intermittent sources such as WTS, DSTS and combination
of both WTS and DSTS is checked.

The objectives of the paper are as follows:

(1) To develop a two area multi source model comprising of thermal and hydro units with GRC and GDB as non-
(2) To compare the performance of PID and PIDN as secondary controller and find the best among them.
(3) To incorporate the distributed generation unit in Area 1 comprising of fuel cell, aqua electrolyzer, Battery
energy storage system and diesel engine generator and study the system dynamics with the best controller
obtained in (2).
(4) To include intermittent sources in the DG unit such as (i) WTS and DSTS one by one and (ii) WTS and DSTS
both together and investigate the system dynamics with the previous case.


An unequal area Hydro-Thermal system is considered here for investigation. Each area has a reheat thermal and a
hydro plant. Capacity of Area 1 and Area 2 are 2000 MW and 4000 MW respectively. The reheat thermal and hydro
units are provided with GDB of 0.06% and, appropriate GRC of 3%/min for thermal and 270%/min and 360%/min
raising and lowering command of a hydro plant is provided. For hydro units, a GRC of 270%/min and 360%/min for
increasing and lowering command is provided. The transfer function model of the proposed system is shown in
Figure 1(a) and Figure 1(b). The nominal parameters of thermal, hydro and DG units are considered from [1], [2],
[9] and [12] respectively. The system dynamic performance is investigated SFS optimized classical controllers and
with 1% SLP at Area 1. Integral time absolute error (ITAE) is considered as performance index with an objective to
minimize the settling time and peak deviations i.e. area control error (ACE) and is given by eqn. (1)
ACE J ITAE ³ ΔF1  ΔF2  ΔPtieerror .tdt (1)

FIGURE 1(a) Transfer function modeling of a two area thermal-hydro system incorporating DG unit in Area 1

FIGURE 1(b) Transfer function modeling of the FIGURE 2 Transfer function modeling of a PIDN
DG unit in Area 1 controller

2.1 Components as Distributed Generation

The components and their brief description are given as follows:

a) Aqua electrolyzer (AE). It is used as it absorbs fluctuating power output from wind and solar, which is
converted to hydrogen and stored for later use as fuel.
b) Fuel cell (FC). It produces power by electrochemical reaction amongst hydrogen and oxygen, until fuel is being
supplied continuously from outside. It contains no moving part, allowing more reliable and quieter
performance. Some of the advantages of FC over conventional generating systems are low environmental
pollution, efficient power generation, variety of fuels, reusability of exhaust heat, and faster installation.
c) Diesel engine generator (DEG). It yields torque which drives the synchronous machine, generating electrical
power output. It is very much necessary that the diesel-prime mover has a speedy dynamic response and nice
ability to reject disturbances as load demanded by consumers undergoes sudden changes [9-10].Wind turbine
system (WTS). The power output from WTS is dependent on wind speed at that particular instant. The pitch
system of WTS sets the pitch angle according to wind speed introducing the nonlinearity.
d) Frequent fluctuations in power from dish-Stirling solar thermal plant may cause large problems i.e. instability in
power systems operation. Installation of Energy storage device can be a viable solution. Due to convincing
features (large energy density, fast access time) in terms of technical and economics, the battery energy storage

system can be termed as sufficient energy storage technology to store large amount of energy [13]. They can
instantly offer the power system with a large amount of the power. Also, a greater amount of energy for a longer
period can be provided with the same. The Battery Energy Storage System (BESS) comprises of a battery bank
and a power converter that interfaces the battery bank to the power grid. Power capacity can be increased by
increasing the number of modules of BESS.
e) Wind turbine system (WTS). The power output from WTS is dependent on wind speed at that particular instant.
The pitch system of WTS sets the pitch angle according to wind speed introducing the nonlinearity.
f) The first-order transfer function model for DSTS is considered from Ref. [9]. The dish-Stirling engine power
output modeling using Beale formula which is a function of the mean cycle pressure, swept volume, cycle
frequency, and Beale number whose typical value ranges from 0.011 to 0.015 and is dependent on heating
source temperature. Solar Insolation or irradiance is dependent on dish Stirling engine power.

2.2 Design of controllers

Classical controllers such as proportional-integral-derivative (PID) and proportional-integral-derivative with first

order filter effect (PIDN) controller are utilized one by one for reducing the peak deviations of the perturbed system
[15].The transfer functions for PID and PIDN controller is given by eqn. (2) and eqn. (3) respectively.

K I (s)
C (s) K P ( s)   suKD (2)

K I (s) Ns
C (s) K P (s)   KD u (3)
s ( N  s)

The considered controller parameters of PID and PIDN are tuned simultaneously by a powerful meta heuristic
algorithm known as Stochastic Fractal Search (SFS) algorithm. The proposal of SFS technique [16] to discover an
area based on the property of a random object which resembles self-similarity of the patterns in nature, such as trees,
snowflakes, animal coloration patterns, crystals, Romanesco broccoli, lungs, river networks, blood vessels and
DNA. SFS accelerates up convergence and accuracy by exchanging information among all participants in the group.
SFS algorithm [16] constitutes of diffusing and two-staged updating process which offers a proper balance of
exploration and exploitation. The flowchart and pseudo code can be referred from [16]. The performance of
heuristic algorithms is problem specific. In this work, the tuned value of parameters of the SFS technique is
considered as population size (PS) = 60, maximum generation (MG) = 50, maximum diffusion number (MDN) = 2
and diffusion walk (DW) = 0.75.


Investigations are carried out in the two area multi source model with and without distributed generation in Area 1.
Without the distributed generation in Area 1, only two thermal and a thermal and hydro plant exist in Area 1 and
Area 2 respectively. The case studies undertaken are as follows:
(1) Controller comparison in absence of DG unit in Area 1.
(2) Incorporating DG unit in Area 1 (without DSTS and WTS unit in it).
(3) Incorporating DSTS plant (having step input) in DG unit in Area 1.
(4) Incorporating WTS plant in DG unit in Area 1.
(5) Incorporating both DSTS and WTS plant (having intermittent input) in DG unit in Area 1.

Case I: Performance comparison of classical controllers for the test system

The test system is provided with step load perturbation of 1% at Area 1. Classical controllers such as PID and PID
with first order filter (PIDN) are employed simultaneously as the secondary controllers in both areas. Effective
tuning of the controller parameters are carried out by SFS algorithm. The optimum controller parameters are noted

down in TABLE 1. The system dynamic responses for each controller are obtained and compared and are presented
in Figure 1.

TABLE 1 Optimum PID and PIDN controller gains and parameters

PID controller gains
KP1* KI1* KD1* N1* KP2* KI2* KD2* N2*
0.058 0.634 0.303 **** 0.343 0.155 0.731 ****
PIDN controller gains
KP1* KI1* KD1* N1* KP2* KI2* KD2* N2*
0.727 0.583 0.646 33.19 0.408 0.564 0.523 58.09



FIGURE 3 Comparison of system dynamics for PID and PIDN controller

(a) ∆F1(t) vs. f(t); ∆F2(t) vs. f(t)’
(b) ∆P g11(t) vs. f(t); ∆P g22t) vs. f(t);

The frequency deviations in both areas along with power generation in both areas for each unit are shown. In the
figure for frequency deviation, it is clear that PIDN controller offers lesser peak deviations and settling time as
compared to PID controller. The responses of power generation are oscillatory in case of PID controller whereas
PIDN offers a less oscillatory and desired output generation. Thus, it can be inferred that PIDN controller is superior
to PID controller for its effective tuning capability and reduction of area control error. Thus in the further case
studies, PIDN controller is utilized as secondary controller in both areas.

Case II: Addition of a DG unit in Area 1.

The test system is now included with distributed generation in Area 1 comprising of Aqua Electrolyzer, fuel cell,
diesel engine generator and Redox flow battery system. The gains and time constants of each unit in DG is presented
in Appendix Section. PIDN controller parameters are now tuned with SFS algorithm and the optimum parameters

are tabulated in column 2 of TABLE 2. The system dynamic responses are plotted and compared with the
corresponding responses obtained when DG unit is not incorporated and, is presented in Figure 4. The system
dynamic responses are seen to have slightly lesser peak deviations (less than 0.01 Hz for frequency deviation) and
settling time than the previous case where the DG unit was not utilized. PIDN controller satisfactorily improved the
system dynamics in presence of DG unit in Area 1.



FIGURE 4 Comparison of system dynamics for PIDN controller with incorporation of DG unit at Area 1
(a) ∆F1(t) vs. f(t)
(b) ∆P tie12(t) vs. f(t).

TABLE 2 Optimum controller gains of PIDN controller under different scenarios from Case II to Case V
Gains Case II Case III Case IV Case V
KP1* 0.749 0.551 0.297 0.893
KI1* 0.527 0.997 1.000 0.394
KD1* 0.824 0.455 0.615 0.972
N1* 38.21 31.9 38.10 30.16
KP2* 0.530 0.368 0.398 0.903
KI2* 0.725 0.076 0.383 0.788
KD2* 0.342 0.326 0.507 0.768
N2* 73.85 40.68 21.77 4.989

Case III: Incorporating DSTS plant in DG unit in Area 1.

In the previous case study, it is observed that inclusion of DG unit slightly improves the system dynamics with that
of the responses obtained without incorporating DSTS plant. DSTS plant, a random power output producer is added
newly whereas Aqua Electrolyzer, fuel cell, DEG, RFB was already present which provides a constant power
output. The power output pattern for the DSTS plant is shown in Figure 5(b). The PIDN controller gains are again
tuned with SFS algorithm and the optimum controller gains are noted in column 3 of TABLE 2.



FIGURE 5 Comparison of system dynamics with and without incorporation of DSTS in DG unit at Area 1
(a) ∆F1(t) vs. f(t)
(b) ∆P g11(t) vs. f(t); ∆P g22(t) vs. f(t); ∆P gDSTS(t) vs. f(t).

The system dynamic responses utilizing DSTS plant in DG unit is being compared with the corresponding response
without using DG unit in it. The resulting figures are depicted in Figure 5. The frequency deviation is not deviating
too much from the situation when DSTS were not installed (same as previous case i.e. 0.009 Hz). The random nature
of power output in Area 1 is due to the presence of DSTS plant. The power output by DSTS is subtracted from the
disturbed power output and the remaining is shared between the thermal and hydro generators.

Case IV: Incorporating WTS plant (having step input) in DG unit in Area 1.
It is observed that with incorporation of WTS plant in the DG unit, the undershoot has increased slightly (>0.0009
Hz) than the previous case is concerned and the system remains stable with the same settling time as in earlier case.
However, the settling time remains the same in both the cases. The changes at 20s, 60s are due to the disturbances of
WTS unit. It is expected that with inclusion of both intermittent WTS and DSTS in the DG unit, the system
dynamics may deteriorate further.


FIGURE 6 Comparison of system dynamics with and without incorporation of WTS in DG unit at Area 1
(a) ∆F1(t) vs. f(t)
(b) ∆P g11(t) vs. f(t); ∆P g22(t) vs. f(t).

Case V: Incorporating both WTS and DSTS plant in DG unit in Area 1.

The below depicted figure presents the comparison of system dynamic responses between responses incorporating
both WTS and DSTS plant in Area 1 to that of responses without incorporating DG unit in Area 1. The undershoot
has decreased slightly and settling time remains almost the same. DG unit has shared a significant amount of power
generation along with thermal and hydro plants.




FIGURE 7 Comparison of system dynamics with incorporation of combination of DSTS and WTS in DG unit at Area 1
(a) ∆F1(t) vs. f(t)
(b) ∆P g11(t) vs. f(t); ∆P g22(t) vs. f(t);
(c) ∆P tie12(t) vs. f(t).

The power generation by each unit matches the generation profile equivalent to the disturbance in the system for
each area. As, there is no disturbance present in Area 2, the change in power generation is shown as zero.

Automatic generation control of a hybrid power system consisting of hydro and thermal plants is considered under
investigation. Distributed generation unit comprising of Aqua Electrolyzer, fuel cell, diesel engine generator and
Redox flow battery is introduced in the Area 1 and frequency and tie power deviations are observed due to step load
perturbation. The secondary controller selection is done between the classical controllers such as PID and PID with
first order filter effect (PIDN). Comparison of performance shows the superiority of SFS optimized PIDN controller
for the two area multisource system. Several case studies with incorporation of Distributed generation show that the
frequency and power deviation decreases (0.01 Hz to 0.008Hz; improvement of 2% in case of frequency deviation)
when Aqua Electrolyzer, fuel cell and diesel engine generator and Redox flow battery is chosen. However, the
improvement in tie-power deviation also remains more or less the same. Further inclusion of intermittent sources
such as WTS, DSTS and combination of WTS and DSTS slightly increases the deviation but the system remains
stable with safe limits of frequency and power deviation. Thus, an extensive study incorporating a mix of both
conventional and intermittent sources as DG in Area 1 is successfully executed with PIDN as secondary controller.


Authors wish to thank the Department of Electrical Engineering, Girijananda Chowdhury Institute of Management
and Technology Guwahati, for providing the necessary facilities for completing this work.


[1] A. Saha, and L. C. Saikia., Performance analysis of combination of ultra-capacitor and superconducting
magnetic energy storage in a thermal-gas AGC system with utilization of whale optimization algorithm
optimized cascade controller (Journal of Renewable and Sustainable Energy, 10, 2018), pp. 014103 (1 - 26).
[2] I. Pan and S. Das, Fractional Order AGC for Distributed Energy Resources Using Robust Optimization (IEEE
transactions on smart grid, 7, 2016), pp. 2175 - 2186.
[3] I. Pan and S. Das, Fractional order fuzzy control of hybrid power system with renewable generation using
chaotic PSO, (ISA Transactions, 62, 2016), pp.19 – 29.
[4] Ting-Chia Ou, Kai-Hung Lu, and Chiou-Jye Huang., Improvement of Transient Stability in a Hybrid Power
Multi-System Using a Designed NIDC, (Energies, 10, 2017), pp.1 – 16.
[5] Ting-Chia Oua, Chih-Ming Hongb, Dynamic operation and control of microgrid hybrid power systems (Energy,
66,2014), pp. 314 - 323.
[6] S. Funcke, D. Bauknecht, Typology of centralised and decentralised visions for electricity infrastructure,
(Utilities Policy, 40, 2016), pp. 67 - 74.
[7] Peperman et al., Distributed generation: definition, benefits and issues, (Energy Policy, 33, 2005), pp. 787–798.
[8] L. Mehigan, J.P. Deane, B.P.Ó. Gallachóir, V. Bertsch, A Review of the role of Distributed Generation (DG) in
Future Electricity Systems (Energy, 163, 2018), pp. 822 – 836.
[9] D. C Das, N. Sinha, A.K. Roy, Small signal stability analysis of dish-Stirling solar thermal based autonomous
hybrid energy system, (Electrical Power and Energy Systems, 63, 2014), pp. 485–498.
[10] D. C. Das, A.K. Roy, N. Sinha, GA based frequency controller for solar thermal–diesel–wind hybrid energy
generation/energy storage system. (Electrical Power and Energy Systems, 43, 2012), pp. 262–279.
[11] A. Rahman, L. C. Saikia and N. Sinha, Automatic generation control of an interconnected two-area hybrid
thermal system considering dish-stirling solar thermal and wind turbine system. (Renewable Energy, 105,
2017), pp.41 - 54.
[12] A. Rahman, L. C. Saikia and N. Sinha, .AGC of dish-Stirling solar thermal integrated thermal system with
biogeography based optimised three degree of freedom PID controller, (IET Renew. Power Gener., 10 (8),
2016), pp.1161-1170.
[13] I.A. Chidambaram, B. Paramasivam, Control performance standards based load-frequency controller
considering Redox flow batteries coordinate with interline power flow controller, (Journal of Power Sources,
219, 2012), pp.292 - 304.
[14] R. Shankar, R. Bhushan, K. Chatterjee, Small-signal stability analysis for two-area interconnected power
system with load frequency controller in coordination with FACTS and energy storage device (Ain Shams
Engineering Journal, 7, 2016), pp.603 – 612.
[15] B. Maâmar, M. Rachid, IMC-PID-fractional-order-filter controllers design for integer order systems, (ISA
Transactions, 12, 2015), pp.19-23.
[16] H. Salimi, Stochastic Fractal Search: A powerful metaheuristic algorithm, (Knowledge-Based Systems, 75,
2015), pp.1–18.


Areas Generations Parameters considered

Tg11 = Tg12 = Tg21 = 0.08s;
Tt11 = Tt12 = Tt21 = 0.1s;
Thermal System
Area 1 Kr11 = Kr12 = Kr13 = 5;
Tr11 = Tr12 = Tr13 = 10s.
Area 2 Tg22 = 0.2s; Krh = -5;
Hydro System
Trh = 28.75 s; Kb = -1;
Trb = 0.5
Area 1 Fuel Cell KFC = 0.01; TFC = 4s.
Area 1 Aqua Electrolyzer KAE = 0.002; TAE = 0.5s
Area 1 Diesel Engine Generator KDEG = 1/300; TDEG = 2s
Area 1 Battery Energy Storage
KBESS = -1/300; TBESS = 0.01s
Area 1 Wind Turbine System KWTS = 1; TWTS = 1.5s
Area 1 Dish Stirling Solar
KDSTS = 1; TDSTS = 5s
Thermal System