Impact of dynamic demand response in the load frequency control

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 02 | Feb -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 262
IMPACT OF DYNAMIC DEMAND RESPONSE IN THE LOAD FREQUENCY
CONTROL
P CHANDRASEKHARA1, B PARASURAM2, C VISWANATH3, A SURESHBABU4
1 PG Scholar, Dept. of EEE, B.I.T Institute of Technology,Hindupur Affiliated to JNTUA, A.P, India
2 Associate Professor in Dept. of EEE, , B.I.T Institute of Technology,Hindupur Affiliated to JNTUA, A.P, India
3 Associate Professor in Dept. of EEE, , B.I.T Institute of Technology,Hindupur Affiliated to JNTUA, A.P, India
4 Lecturer in Dept. of EEE, ,Mizan-Tepi University,Tepi,Ethiopia
---------------------------------------------------------------------------***--------------------------------------------------------------------------
Abstract:- Demand Response (DR) has proved to be an
inevitable part of the future Power-grid. Most
research works have been documented in the
literature on the benefits and implementation of DR.
However, some works have been reported on the
impacts of DR on dynamic performance of power
systems, specifically on the load frequency control
(LFC) problem. This paper makes an attempt to fill
this gap by introducing a DR control loop in the
traditional LFC model (called LFC -DR) for a single-
area power system. The model has the feature of
optimal operation through optimal power sharing
between DR and supplementary control. The effect of
DR communication delay in the controller design is
also considered. It is shown that the addition of the
DR control loop increases the stability margin of the
system and DR effectively improves the system
dynamic performance. Simulation studies are carried
out for single area power systems to verify the
effectiveness of the proposed method.
Index Terms—Demand response (DR), linear
quadratic regulator (LQR), sensitivity, single-area
power system model, smart grid, stability, steady-
state error.
I. INTRODUCTION
Traditionally, frequency regulation in power
system is achieved by balancing generation and demand
through load following, i.e., spinning and non-spinning
reserves. The future power grid is foreseen to have high
penetration of renewable energy (RE) power generation,
which can be highly variable. In such cases, energy storage
and responsive loads show great promise for balancing
generation and demand, as they will help to avoid the use
of the traditional generation following schemes, which can
be costly and/or environmentally unfriendly.
Given the limited availability, low efficiency, and
high cost of large storage devices, real time smart
responsive load participation, known as demand response
(DR), has been actively considered for power balancing. It
can be achieved by active consumer participation in real-
time to maintain balance between generation and demand
with two-way communication. It is well known that DR
increases system reliability and flexibility to manage the
variability and uncertainty of some RE resources,
decreases the cost of operation, and enhances system
efficiency.
Furthermore, DR can be used to provide ancillary
services (AS) for regulation reserve and to respond
momentarily to the area control error (ACE). A number of
studies have also addressed the effectiveness of de-
centralized dynamic demand control on stabilization of
grid frequency, mainly at the transmission level. However,
the above studies present the following shortcomings:
 They do not present a general framework for the
analysis of the impacts of DR on a general power
system model and load.
 AGC model has not been considered in the
analysis.
 Only specific loads (such as HVAC, EWHs and
lighting) have been considered in and specific
power systems without generalization.
 Communication delay in central DR, and
measurement delay in decentralized DR have not
been considered.
 Frequency regulation as AS have not been
studied. Only under-frequency load shedding
(UFLS) characterization has been analysed.
 Unreal assumptions for the availability of DR at
all times have been made.
 Load-damping coefficient, which can improve
frequency stabilization, has been ignored.
 Only sensitivity analysis of frequency-related
load-damping coefficient characteristic without
generalization and DR control is presented.
In the last ten decades, traditional LFC models have
been revised and modified to include the different types of
power plants, including RE power generation with actual
limitations, such as ramp-up/down limits, in the
traditional and deregulated power market. These models
are useful in small disturbance studies such as small
variations in load and generation, and in controller design.
However, so far in the literature, the extreme shedding of
the responsive loads in emergency DR could lead to
unexpected power oscillations, which complicate the

sequential generation control. The LFC -DR model will
help the concept of control in the LFC model has only
focused on the generation side, and DR has not been
included in these studies. In this paper we modify the
general small-signal model of a power system used in LFC
studies by introducing a DR control loop to the LFC model
(called LFC-DR). Other goals of the paper are to make the
model as general as possible devices. This is an important
parameter in the system dynamic performance of LFC-DR.
We have assumed the communication delay between the
balancing authority (BA) and the Lagcos to be the same as
that between the BA and generation companies (Gencos).
We have not considered these delays in our study since
the focus of this paper is on the evaluation of the DR loop
in the LFC model. The proposed LFC- DR also gives an
opportunity to the system operator to choose the DR
option or spinning/non-spinning reserve, or a
combination of the two, based on the real-time market
price. Furthermore, the LFC-DR model can be used to
estimate the actual value of the required responsive load
manipulation when the magnitude of the disturbance is
unknown to the system operator.
The idea of DR for AS used in this paper, has been
fully explored in our previous work and will not be
repeated here. In such a model, the Lagcos will work with
the customers and inform the utilities, e.g., independent
system operators, of the amount of DR available. An
example of such a model is the PJM electricity market.
The remainder of this paper is organized as follows:
In Section II, the concept of LFC-DR model is developed
for a single-area power system. The model is analytically
evaluated in Section III. The controller design is presented
in Section IV, and simulation results are given in Section V.
Section VI includes a discussion of exceptional cases not
explored in this paper and the future work. Finally, the
paper is concluded in Section VII and to include
communication latency associated with DR between the
load aggregator companies (Lagcos) and the end use
customers.
II. PROBLEM FORMULATION FOR SINGLE-AREA
POWER SYSTEM
The general low-order linearization power
system model for the purpose of frequency control
synthesis and analysis is given by the power balance
equation in the frequency-domain:
( ) ( ) ( ) ( )-------(1)
( ) ( )-internal power mismatch;
( )---- Frequency deviation;
2H ------- equivalent inertia constant;
D-equivalent load-damping coefficient;
Fig-1 Block-diagram representation of a single-area
power system model.
Since DR for AS performs like spinning reserve in
magnitude and power flow direction, i.e., once frequency
deviation is negative (positive), it is required to turn OFF
(ON) a portion of the responsive loads (i.e., DR), (1) can be
simply modified as follows to include DR
( ) ( ) ( ) ( ) ( ) …2
In some earlier works such as the effect of DR has
been included in the load-damping coefficient, D. We
believe the effect of DR should be separated because D is
an inherent parameter of the system and is not a
controllable one, whereas DR is an intentional control
signal. In addition, (2) will permit to have a separate
control loop for DR, which is more realistic and gives a
better structure for controller design. The block diagram
for single area power system with a simplified non-reheat
steam turbine is shown in Fig. 1, where the feed-back loop
for DR is also shown. Tg and Tt are the equivalent speed-
governor and turbine time constants, respectively, R is the
equivalent droop value, and Td is the equivalent DR delay.
The parameters of the system can be the equivalent of all
generation assets and load damping of the same area. This
model is selected to convey the main idea of this paper. In
our future work, this model will be extended to multi-area
interconnected Power-systems with multiple Gencos and
Lagcos.
Unlike the usual spinning reserve-provider
power plants, there is no ramp up and down limitations
on the DR resources. the only obstacle for DR
(disregarding the aggregation of small loads) is
communication delay, known as latency, which could
affect the system dynamic performance.
A. State-Space Dynamic Model for LFC-DR
State-space representation of the LFC model is a
useful tool for the application of modern/robust control
theory. It can also be used for creating a general
framework of LFC in dynamic frequency analysis which
can be conveniently modified and applied to power
system of any size. We therefore derive the dynamic
model of the power system, including DR in the state-

space representation, to study the effect of DR on LFC
performance and controller design. Although the
proposed LFC-DR model of Fig. 1 is based on a simplified
power system model with a non-reheat steam turbine,
similar analysis and conclusions can be extended to other
types of turbines, such as hydro or reheat-steam turbine.
The state-space realization of a single-area power
system with DR (shown in Fig. 1) is given by
( ) ( ) ( )------ (3)
y(t)=C.X(t)
Where A is the system matrix, B is the control input
matrix, is the disturbance matrix, X is the state vector,
u(t) is the input vector, w(t) is the disturbance variable, C
is the observation matrix, and Y(t) is the system output. In
order to derive the linear state-space model of the system,
it is required to have a linear model of the system under
study. From Fig. 1, it can be seen that the system has only
one nonlinear element which is the time delay in the DR
control loop. Therefore, we need to linearize the time
delay for derivation of the linear state-space model.
Padeapproximation, used for linearizing the DR time
delay, is explained in the following subsection.
B. Pade Approximation
Padeapproximation has been widely used to
linearize systems with time delays in control engineering
with very strong convergent results. It basically
approximates time delays by a quotient of polynomials.
Specifically, the Pade function for e-sTd = Rpq(-sTd) defined
as follows:
where
pq and pq are polynomials of order and
respectively. It is usually common for the numerator and
denominator of the approximation fractional functions to
have the same order, and the order usually varies
between 5 and 10. Fig. 2 shows the phase of the step
response in frequency-domain of the Pade approximation
with different orders in comparison to a pure time delay
of 0.1 sec. Since the cut-off frequency of the low pass
filters, i.e., speed-governor and turbine, in the model of
the power system are usually less than 15 rad/sec, the
5th-order Pade approximation is acceptable and is used in
this study. The magnitudes of all orders of
padeapproximation in the frequency domain have also
been compared to that of pure time delay. They are all at 0
dB and not shown here. The signal flow graph for the
state-space model of the power system shown in Fig. 1
with 5th-order padeapproximation is sketched in Fig. 3.
The state variables are also shown in this figure for future
reference. The gains of the feedback and feed-forward
paths of the 5th-order padeapproximation are given in
Table I. Td is the DR communication latency.
Fig-2. Different order of padeapproximation for pure time
delay Td=0.1 sec,.
Fig-3. Signal flow graph of the single-area power system
model with 5th-order padeapproximation.
TABLE-1
Padeapproximations Feed-Forward and Feedback Values
for Signal Flow Graph
With the above approximation of time delay nonlinearity,
the state-space representation of the system (3) has the
matrices given in (7)–(10), where is the transpose
operation of matrices. For more complicated power
systems, the upper left partition of matrix A, and the left
partitions of other matrices can be modified to represent
the new power system model. power system model.

The other partitions of the matrices should be properly
resized based on the order of the new
III. ANALYTICAL EVALUATION OF THE MODEL
In this section, steady-state error evaluation,
sensitivity analysis, and system stability of the LFC model
with and without the DR control loop are presented.
A. Steady-State Error Evaluation
The primary control loop in Fig. 1, known as
frequency droop control, is the fastest intentional control
action in a power system but it is not enough to make the
frequency deviation go to zero at steady-state. For this
reason, the supplementary frequency control loop, as
shown in Fig. 1, is required for further control. However,
the DR control loop is also added to the problem in this
study. Therefore, it is necessary to investigate the impact
of the DR control loop on the steady-state error of the
given power system in Fig. 1. Later in this subsection, a
synthesis of controller design, based on optimal sharing
between DR and supplementary control loops, will be
derived from the steady-state error evaluation. The
conventional LFC steady-state equations are well-
documented. Adding the DR control loop to the
conventional LFC model, the system frequency deviation
can be expressed as follows:
( )
( )
( ( ) ( ) ( ) ( ))
( ) ( ) ( ) ( )
It can be
seen that any type of power system model with equivalent
turbine and governor can be represented by modifying
H(s). Substituting (12) into (11) yields
( )
( )
( ) ( ) ( ) ( ) ( ) ( ))
Solving (14) for f(s) will result in the frequency deviation
equation as follows:
In the LFC analysis, it is common to use a step load
disturbance ( ) for as
( )
Based on the final value theorem, and substituting (17)
into (15), the steady-state value of the system frequency
deviation can be obtained as follows:
Therefore, is equivalent to the system frequency response
characteristics, B, and the steady-state frequency
deviation can be written as follows:
It can be seen from (22) that the frequency deviation will
not be zero unless the supplementary and/or DR controls
exist. Also, DR control loop gives an extra degree of
freedom for system frequency regulation. In addition, the
following conclusions can be drawn from (22):
• The steady-state error is not dependent on the delay and
the order of its approximation.
• With DR available in the LFC, a higher reliability of
frequency regulation can be achieved, since the DR
control loop can complement the supplementary control
loop. In cases when the supplementary control is not
available, the performance of the frequency regulation can
be guaranteed by the DR loop, if enough DR resources are
available.
• In order to have zero frequency deviation at steady-
state, the required control effort can be split between the
supplementary and DR control loops. In other words, an
ISO/RTO will have the opportunity to perform the
regulation services in a cost effective way and analyze the
frequency response of the system quickly. This goal can be
achieved only in the proposed formulation with an added
control loop for DR (Fig. 1).
Further discussion to the last conclusion is:
consider a situation where there is no DR available. The
frequency error will be zero at steady-state-if It means
that the supplementary control should provide enough
spinning and/or non-spinning reserve at the time of
disturbance.
( ) . With DR available in the LFC, the required
control effort, called in this study, can be split between the
two controls loops based on their cost at real-time
electricity market as follows:
( )
( ) ( )
Where 0< α<1 is the share of traditional
regulation services in the required control effort.
means that the total required regulation will be provided
by the traditional regulation services, i.e. spinning and
non-spinning reserve, and is for the time when all the
required control would be provided by DR. it is possible
for the ISO/RTO to effectively and quickly assess the

different scenarios of LFC to evaluate the system
performance under various circumstances. Finally, the
steady-state value of the two inputs should be
( )
( ) ( )
B. Sensitivity Analysis for the Feedback System With
and Without DR
In this subsection, an analytical method is utilized
to study the impact of the DR control loop on the overall
sensitivity of the closed-loop system w.r.t. the open-loop
system. It is also desired to measure the sensitivity of the
closed-loop system w.r.t. the coefficient. The first
sensitivity analysis is quite important since it shows the
robustness of the closed-loop system performance when
the system parameters are subjected to any change or
variation.
The second sensitivity analysis is also necessary
since α is an important parameter in the performance of
the LFC-DR model.
Fig-4. Modified power system model with integral
controller for DR and supplementary control loops.
The power system model shown in Fig. 1 is
modified for this part with a single integral controller
(with gain) for both the supplementary and DR control
loops, and also for the rest of this paper, as shown in Fig.
4. This modification will also allow us to split the required
control effort between the DR and supplementary control
loops, as was discussed in Section III-A.
For the power system shown in Fig. 4, the closed-
loop transfer function relating the system frequency
deviation to a step change in the load can be derived as
follows:
In (25), the first expression is the closed-loop
transfer function when both DR and supplementary
control loops are available, whereas the second equation
shows the closed-loop transfer function for conventional
LFC (with no DR).
Where K is integral feedback gain of the system and M(s)
is
( )
( )
The open-loop transfer function, where only the primary
control exists, can be derived as follows:
In order to derive the sensitivity function of the
closed-loop system w.r.t. the open-loop system, (25) can
be simplified and rearranged using (27).
Fig-5. Closed-loop system w.r.t. the open-loop system
sensitivity values of a simulation study for LFC with and
without DR.
TABLE II
POWER SYSTEM PARAMETERS FOR THE SIMULATION
STUDY
Therefore, the unit less sensitivity function of the
closed-loop system w.r.t. the open-loop system, for
systems with and without DR, can be written as follows:
It is noted from (29) that the closed-loop system
is highly sensitive to the changes in the open-loop system,
i.e., any change in the value of TOL will have a large effect
on SDR
OL and SS
OL. From (29), the ratio of sensitivity
functions can be expressed as
Two different values for are used in Fig. 5 to show
that a higher DR share (smaller) will result in less

sensitivity of the closed-loop system w.r.t the open-loop
system. That is, the closed-loop system becomes less
sensitive to the variation of uncertain parameters of the
open-loop system. When (i.e., 80%of the required
regulation would be provided by the supplementary
control and 20% is from DR), the sensitivity values for
both the closed-loop systems, with and without DR, are
almost similar. This is a good indication that the 5th-order
Padé approximation, used for linearizing the time delay in
the DR loop, doesn’t have any negative impact on the
system performance. A similar study has been carried out
to investigate the sensitivity of the closed-loop system
w.r.t. the integral feedback gain, and similar results have
been obtained. Since is an important parameter in the
performance of the closed-loop system with DR, it is
useful to evaluate the sensitivity of the closed-loop system
w.r.t. this coefficient. The sensitivity function can be
written as follows:
The previous simulation setup was utilized for two
different values of Sensitivity results are shown in Fig. 6.
Fig- 6. Closed-loop system w.r.t. sensitivity values for the
LFC-DR model, α=0.1andα=0.8.
It can be seen from Fig. 6 that the closed-loop
system is less sensitive to when the DR control loop takes
a higher share in the frequency regulation, i.e., smaller
values.
C. Stability Analysis of the Closed-Loop Systems
A prerequisite for satisfactory control of a
feedback control system is its stability, and gain and phase
margins are two criteria commonly used for stability
evaluation. The gain and phase margins can be obtained
from the Bode diagrams of the open-loop transfer
function of the closed-loop system since the zeros of the
characteristic equation are poles of the closed-loop
system. Using the load disturbance, as the system input,
the open-loop transfer functions are
where ϑS and ϑDR are the open-loop transfer function for
the closed-loop systems with and without DR, respectively
IV.GENERAL APPROACH FOR CONTROLLER DESIGN
FOR THE LFC-DR MODEL
Several different classical and modern control
theories have been utilized for the LFC problem. In this
section, a general controller design approach for the LFC
problem with the DR control loop is presented. Here, we
use the widely used linear quadratic regulator (LQR) for
controller design. We will examine other effective control
design approaches, which will include parameter
uncertainty and measurement noise, in our future work.
The simplified version of the LQR problem is to
design the controller such that the performance index
(35) is minimized for the system given in
(35)
Where ϱ is a weighting factor chosen by the designer,
considering the trade-off between system transient
performance and control effort. Q is an n-by-n semi-
definite symmetric state cost matrix ( is the number of
system states), Q is an n-by-n positive definite symmetric
control cost matrix (m is the number of control
inputs),XT=[X1,X2, X3….] and (where X1=∆f,X2=∆Pg, X3=∆P
and X4,X5, X6 are the states associated with the 5th-order
Padéapproximation). Using (23), it can be shown that the
control input to the supplementary-controller (u1=∆Ps)-
and (u2=∆PDR)to the DR controller are related as follows:
For the system with unified control inputs, all the state-
space matrices will remain unchanged, except the control
input matrix, where (8) is modified to include
(u2=F(u1))as follows:
It is also possible to use u1=F(u2) in (8). However, as will
be shown in the simulation results in Section V, the
system performances in the two cases are nearly identical.
It is also noted that all the system states are non-zero
except ∆f, where the goal is to keep ∆f as close to zero as
possible. As a result, the full-state feedback controller
cannot guarantee zero steady-state error for frequency
deviation. Therefore, an integral controller is necessary to
ensure zero steady-state error in the system frequency. In

the early works on the application of optimal control to
the LFC problems, it was common to fix this problem with
redefining the states in terms of their steady-state values,
which essentially needs a prior knowledge of the
disturbance. However, in most real-world cases the
disturbance to the power system is an unknown
parameter. Therefore, in this study, an adhoc solution to
the integral control problem is utilized by augmenting the
state vector. The design will also be robust against any
changes in the system parameters. Modifying the state-
space model in (3) to include the integrator, the
augmented state equations become
Where ữ=u2. The augmented state-space equation can be
written as follows:
(39)
where the states are defined as XT=[∆f, ∆Pg, ∆Pt, X4,X5, X6.
The matrices for the modified system are
If the augmented system matrix is controllable, then the
control law and the state feedback can be defined as
To employ the LQR method, it is required to
define the state and control weighting matrices, Q and R
(scalar quantity), respectively. Before defining the
weighting matrices, a set of frequency response
requirements should be defined:
• The steady-state frequency deviation following a step-
change in the load must be zero, i.e.[∆(f)]2,
• The time error represented by the integral of the
frequency deviation should not exceed 3 seconds, i.e. [∫
∆(f).dt]2,
Eventually, the weighting matrices considering the above
requirements will be as follows:
Q=diag(1 01x7 1) and R=[ 1 ] ……(42)
In the next section, simulation results for the LFC-DR
model of a single-area power system are presented to
verify the effectiveness of the proposed model.
V. SIMULATION RESULTS FOR SINGLE-AREA POWER
SYSTEM
In order to show some important features of the
proposed LFC-DR model, the results of several different
simulation studies are reported in this section for a single
-area power system. In order to make a fair comparison,
similar LQR design procedure has been employed for
controller design for both systems, with and without DR.
The parameters used in the simulation studies are given
in Table II. The LQR problem has been solved using
MATLAB/Control System Toolbox.
In the first simulation study, a 0.01 pu load
disturbance was applied to the single-area power system
with conventional LFC and the proposed LFC -DR model.
The system frequency deviation is shown in Fig. 8.
Fig- 8. Frequency deviation for conventional LFC and LFC-
DR models.
Fig-9. Steady-state values of the control inputs for the
LFC-DR model
It can be seen that when 0.1 (i.e., 10% of the
required regulation is provided by the supplementary
control and 90% from DR), the LFC-DR model has a
superior performance over the conventional LFC during
the transient period. Specifically, the overshoot in the
system frequency deviation is decreased by about 42.5%.
The results show improvement in the settling time as
well. The same simulation was repeated for α 0.8 As
expected, the lower DR control effort resulted in less
improvement in the system dynamic performance. It can
be observed that the dynamic performance of the system
approaches that of conventional LFC for higher values of
α.
The supplementary and DR control inputs are
shown in Fig. 9, for the same simulation. As discussed in
Section III -A, the steady-state values of the control inputs
are based on the share between the DR and the
supplementary control loops, i.e., the value of , which is
decided by the regional ISO/RTO based on the real-time
electricity market. The steady-state value calculations are
also shown in Fig. 9 which match with (24).
A simulation study was carried out to show the
impact of the order of Pade approximation on the
performance of the system, the results of which are shown
in Fig. 10. 2nd- and 5th-order Pade approximations are
0 1 2 3 4 5 6 7 8 9 10
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Time (s)Frequencydeviation,Hz
LFC with Dr, alpha = 0.1
LFC with Dr, alpha = 0.8
Conventional LFC
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
x 10
-3
Time (s)
Controlinputs,p.u
Supplementary, alpha=0.8
SS=0.8x0.01=0.008 p.u
DR loop, alpha=0.1
SS=0.9x0.01=0.009 p.u
DR loop, alpha=0.8
SS=0.2x0.01=0.002 p.uSupplementary, alpha=0.1
SS=0.1x0.01=0.001 p.u

considered in the proposed LFC-DR model and compared
with the conventional LFC, for 0.1 .
It can be seen from Fig. 10 that the results from
the 2nd- and 5th-order Padeapproximation are almost
identical. It is mainly because the simplified governor and
turbine models are low pass filters which restrict the
system response to lower frequency ranges, where Pade
approximation is exactly the same as pure time delay, as
discussed in Section II.
Fig- 10. Controller performance for different order of
Padéapproximation.
Therefore, for simplicity, 2nd-order
Padeapproximation can be employed for more
complicated power systems without negative impacts on
the final results.
Fig. 11. Impact of different unified inputs on the
performance of the LFC-DR model.
The two control inputs are unified as a single
input for the controller design as a function of α. The
control input unification can be done in two ways:
unifying u1(t) as a function of u2(t) or vice versa
[ ( ) or ( ) ]. To show the impact of
unification, a simulation study was carried out to compare
the performance of the system for both unification cases,
and the results are shown in Fig.11. It can be observed
that the difference between the two unifying approaches
is negligible. In other words, the unifying control input
can be chosen arbitrarily without any negative impact on
the performance of the LFC-DR.
Fig. 12. Impact of latency on the performance of the LFC-
DR model.
One significant feature of the proposed LFC-DR
model is the possibility for the ISO/RTO to evaluate the
impact of communication delay of the DR control loop on
the system performance for frequency stabilization. In
order to show the impact of latency, a simulation study
was performed for different values of communication
latency for α=0.1. Simulation results are shown in Fig.12.
The lowest communication delay (lowest) is for a small
power system with fast two-way communication link,
such as wireless communication, between the Lagcos and
individual loads. It can be seen that the LFC-DR model
gives a better performance compared to the conventional
LFC when . When the time delay exceeds 0.2
sec, it deteriorates the performance of the LFC-DR, and
the response is even worse than that of conventional LFC
for . This is not surprising since the single-
area power system under study has a very fast dynamic
response. In larger power systems with generation rate
limiters and slow turbine-governor systems, a slower
dynamic behavior would be expected from the
supplementary control. But, the LFC-DR will keep its
superior performance even for higher communication
latencies
Fig. 13. Impact of DR latency on the performance of a
slower single-area power system.
Table IV: Power System Parameters for the Simulation
Study
Tg Tt R 2H D ΔPL
0.3
sec
0.8
sec
2.4
Hz/p.u
3.0
pu
sec
0.0083
p.u./Hz
0.01
p.u.
To show the impact of the DR latencies on a
larger power system with high inertia and slower
response, another simulation study was conducted with
the parameters given in Table IV. It can be seen from Fig.
13 that for a larger and consequently slower power
system, the performance of the LFC-DR model is superior
to that of conventional LFC even for larger communication
latencies. It has been shown in that even with the current
Internet infrastructure, a latency of 500msec can be
achieved easily. Therefore, it can be concluded that the DR
with the largest available latency (500msec) still can be
effective for large power systems.
VI. DISCUSSION AND FUTURE WORK
0 1 2 3 4 5 6 7 8 9 10
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Time (s)
Frequencydeviation,Hz
DR 2nd order pade approximation
DR 5th order pade approximation
Conventional LFC
0 2 4 6 8 10 12 14 16 18 20
-14
-12
-10
-8
-6
-4
-2
0
2
x 10
-3
Time (s)
u2 = F(u1)
u1 = F (u2)
0 2 4 6 8 10 12 14 16 18 20
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Times (s)
Td = 0.02 sec
Td = 0.1 sec
Td = 0.2 sec
Td = 0.3 sec
Td = 0.4 sec
Conventional LFC
0 2 4 6 8 10 12 14 16 18 20
-4.5
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
x 10
-3
Time (s)
Td=0.1 sec
Td=0.3 sec
Td=0.5 sec
Conventional LFC

Our proposed LFC- DR model responds to all
frequency deviations as is the case in the traditional LFC
model. However, if it is desired to prevent the LFC -DR
model to respond to small frequency deviations, and also
keep the linearity of the model, a dead-band could be
added to the input, PL could include the variation in any
renewable generation that might be available in the
power system as negative load. This is because of the fast
dynamics of the common variable generation (wind, solar
PV) compared to those of traditional power plants in the
LFC model.
In this paper, we have explored the effectiveness of
the LFC-DR model for frequency regulation at the
transmission level in a single - area power system.
However in general, large power systems are multi-area
where different Gencos and Lagcos are available in each
area. In our future work, we will report the application of
LFC-DR in multi-area power systems.
VII. CONCLUSIONS
In this paper, a general framework is proposed to
include DR into the LFC problem (LFC -DR). The proposed
formulation can be expanded easily for any type of power
system in size and characteristics. The framework adapts
a real-time electricity market with existing load
aggregators. It balances the power between generation
and demand and stabilizes the system frequency by
utilizing a percentage of available controllable loads
and/or conventional supplementary control, based on the
real- time market price. It also includes communication
latencies in DR for controller design, using
Padeapproximation. It is shown through different
analytical studies that the proposed LFC-DR framework
will improve the stability margins in the conventional LFC
model and is slightly less sensitive to the variation in the
system parameters, such as changes in the open-loop
transfer function. Similar results have also been obtained
for the sensitivity of the closed-loop system w.r.t. the
parameter . Finally, the well-known LQR design is
applied for full-state feed-back controller design for a
single-area power system. Simulation results show the
effectiveness of the LFC-DR model in improving
stabilization of the system frequency.
REFERENCES
[1] C. Goldman, M. Reid, R. Levy, and A.
Silverstein, Coordination of Energy
Efﬁciency and Demand Response a
Resource of the National Action Plan for
Energy Efﬁciency, U.S. Environmental
Protection Agency, U.S. Department of
Energy , 2010, Tech. Rep.
[2] S. A. Pourmousavi and M. H. Nehrir,
“Real-time central demand response for
primary frequency regulation in microgrids,” IEEE Trans.
Smart Grid, vol. 3, no. 4, pp. 1988–1996, Dec. 2012.
[3] D. Jay and K. S. Swarup, “Dynamic demand response
and control in smart grid environment,” in Proc. Annual
IEEE India Conf. (IN- DICON), Hyderabad, 2011, pp. 1–4.
[4] Z. Changhong, U. Topcu, and S. H. Low, “Frequency-
based load control in power systems,” in Proc. American
Control Conf. (ACC), Canada, 2012, pp. 4423–4430.
[5] L. R. Chang-Chien, L. N. An, T. W. Lin, and W. J.
Lee,“Incorpo- rating demand response with spinning
reserve to realize ana adaptive fre- fuency restoration
plan for system contingencies,” IEEETrans.Smart Grid, vol.
3, no. 3, pp. 1145–1153, Sep. 2012
BIOGRAPHIES
P.Chandrasekhara recevied his B.Tech (EEE) degree from
JNTUA Anantapur, Andhrapradesh, India in 2011. Interest
in Power System, Solar Systems and Electrical machines
B. Parasuram his B.Tech (EEE) degree from JNTUH,
Hyderabad, Andhra Pradesh, India in 2006, M.Tech
Energy System from JNTUCEA, Anantapur, Andhra
Pradesh, India in 2011. He is currently working as
Lecturer in Assistant Professor in B.I.T Institute of
Technology, A.P, India. Interests in Power System
operation and control, and Power distribution systems.
C.Viswanath received his B.Tech (EEE)
degree from JNTU, Hyderabad, Andhra
Pradesh, India in 2005, and M.Tech in
Advanced Power System from JNTUCE,
Kakinada, and Andhra Pradesh, India in
2007. He is currently working as
Associate Professor in B.I.T Institute of
Technology, A.P, India. His areas of
interest are Power System operation and control, control
systems, Power distribution systems & Distributed
Generation.
A. Suresh Babu received his B.Tech (EEE)
degree from JNTUCE, Anantapur, Andhra
Pradesh, India in 2003, M.Tech in
Advanced Power System from JNTUCE,
Kakinada, Andhra Pradesh, India. He is
currently working as Lecturer in Mizan-
Tepi Univeristy, Tepi, Ethiopia. His areas
of interest are Power System operation and control,
control systems and Power distribution systems.

Impact of dynamic demand response in the load frequency control

More Related Content

What's hot (20)

Similar to Impact of dynamic demand response in the load frequency control (20)

More from IRJET Journal (20)

Recently uploaded (20)

Impact of dynamic demand response in the load frequency control