2014 IEEE 8th International Power Engineering and Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah,

Malaysia. 24-25 March 2014

Under-Frequency Load Shedding Technique

Considering Response Based For Islanding
Distribution Network Connected With Mini Hydro
Hasmaini Mohamad,
Salmah Sahdan, Nofri NY Dahlan

Norazliani Md Sapari
University of Malaya Power Energy
Dedicated Advanced Centre (UMPEDAC),
Level 4, Wisma R&D, 59990,
Kuala Lumpur, Malaysia.

Faculty of Electrical Engineering, University

of Technology MARA(UiTM),
Shah Alam, 40450,Selangor, Malaysia

rapidly and the situation can lead to the system collapse. The
frequency however can be restored by shedding certain
amount of load so that the system is back into equilibrium.
This method is called Under-Frequency Load Shedding
(UFLS).

Abstract One of the biggest challenge in an islanding

operation is to sustain the frequency stability. A large power
imbalance following islanding would cause under-frequency,
hence an appropriate control is required to shed certain
amount of load. Under-frequency load shedding (UFLS)
scheme is commonly adopted for grid transmission system of
which requires particular consideration for application on an
isolated/islanded system. The main objective of this research is
to develop an adaptive UFLS technique for an islanding
system. The technique is designed considering response based
for any disturbance during islanding operation. The
disturbance magnitude is determined from the swing equation
in which relies on frequency and rate of change of frequency
measurement for shedding the required amount of load. The
technique is modeled by using PSCAD simulation tool. A
simulation study on a distribution network connected with
mini hydro generation is carried out to evaluate the UFLS
model. It is performed under different load condition: peak
load, base load and overload scenario. The results have shown
that the load shedding technique have successfully shed certain
amount of load and stabilized the system frequency following
overload.

There are many UFLS techniques that have been

reported. It can be mainly categorized into conventional and
adaptive technique. Conventional is a traditional technique
while adaptive is a dynamic and modern techniques. The
traditional scheme is a simple scheme which shed a certain
amount of load when the system frequency falls below a
certain threshold, and continuously shed until the overload is
relieved. On the other hand, the main concern of adaptive
technique is to shed an optimum amount of load while
restoring the system frequency. This adaptive technique can
be used in both scenarios which are event and response based
[2]. In event based, the algorithm will calculate the total
power imbalances between generation and total load
demand. The load will be shed according to the ranked load.
The frequency response of an isolated and islanded
system could be more severe when subjected to a transient
disturbance. The severity greatly varies with the size of the
isolated system or in other words the total system inertia.
The smaller system inertia tends to cause a quick frequency
drop. It also has a limited spinning reserve to cover for the
shortfall of the generation. Thus, different approaches of
load shedding scheme is needed in designing the adaptive
load shedding scheme in the isolated system [3-6]. For the
islanded distribution system, not much work has been
carried out. The developed load shedding schemes so far
have focused on the approaches for obtaining the optimal
load shedding in the islanded system. Among the strategies
are the scheme based on the frequency and df/dt
information, customers willingness to pay and load
histories [7] and the best time to shed the loads [8].

Keywords-Islanding; Under-Frequency Load Shedding

(UFLS); Islanding Operation; Rate Of Change Of Frequency
(ROCOF); Distributed Generator (DG);

I.

INTRODUCTION

The rapid growth of Distributed Generation based on

renewable energy and distributed technologies such as solar,
hydro, fuel cells, microturbines and energy storage connected
directly to the distribution system contribute to the increased
in generation capacity. One of the main benefit of DG is to
improve the reliability and security of power supply through
the implementation of an islanding [1].
Islanding is defined as a part of distribution system
disconnected from main supply or grid, while the load in that
section is fully supplied by DG. Islanding could be formed
when there is a loss of main or fault occurs in the distribution
system. When system is islanded, DG will solely dispatch
power to the load in the island. A sudden change in
generation over load following system become islanded can
result in the deviation of frequency response during
islanding. A large power imbalance with excessive load will
lead to under frequency. The drop in frequency can occur

A new load shedding technique for an islanding

operation has been presented in [9]. It introduced an
adaptive load shedding considering event and response
based for islanding. This paper extends the study by
evaluating the technique in a different islanding network
with a different range of load. This paper only considers

978-1-4799-2422-6/14/$31.00 2014 IEEE

488

2014 IEEE 8th International Power Engineering and Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah,
Malaysia. 24-25 March 2014

The algorithm of Load Shedding Controller Module

(LSCM) as shown in Figure 2 is the main function of UFLS
scheme. The algorithm always checks whether the system is
islanded. The ROCOF and breaker status at the grid and
generator is continuously monitored by LSCM. The LSCM
can detect any occurrence of disturbance and subsequently
activates the load shedding algorithm. The LSCM will
determine the amount of load that need to be shed based on
the amount of power imbalance.

response based during islanding operation. The load

shedding scheme manages the amount of load to be shed
based on disturbance magnitude/power imbalance. The
amount of disturbance magnitude is estimated based on the
swing equation or specifically the ROCOF technique. The
technique optimally shed the load in accordance with the
load priority. The effectiveness of the technique is
investigated by performing a dynamic simulation on
distribution network that consists of a small unit of hydro
generation operating in an islanded mode. Several case
scenarios are simulated by using PSCAD/EMTDC software.
This paper comprises of four sections including the
current section. The concept of the UFLS technique and the
test system are described in section II. Section III presents
the result and discussion of various test cases. Finally, in
Section IV, conclusion is presented.
II.

UNDER FREQUENCY LOAD SHEDDING SCHEME

The proposed technique introduces an adaptive UFLS by

using response-based method for an islanding operation.
Response based considers any event or disturbance during
islanding operation. Different approach of UFLS to address
the issue of frequency drop during islanding is needed. It
needs an immediate action by shedding some of the load
following disturbance event. This technique uses rate of
change of frequency (ROCOF) and power imbalance to
determine the amount of load to be shed. Figure 1 shows the
concept of UFLS technique applied in this study. A new
module referred as Load Shedding Controller Module
(LSCM) is introduced. Figure 2 shows in detail the
algorithm developed for the LSCM. The module will
response in accordance to the system frequency response.
For safe operation, the system frequency must be on the
specified range. If the frequency falls exceeding the
threshold limits (47.5Hz), the protection system will trigger,
thus the islanding operation could not be continued.

Figure 2: Load Shedding Controller Module (LSCM)

The load shedding controller (LSCM) is modeled to

calculate the power imbalance by using swing equation. The
equation is expressed as in equation (1) :



A. Modelling Load Shedding Controller (LSCM)

H
fn
dfc
Pm
Pe

= The generator inertia constant in second.

= The rated value of frequency in Hertz.
= The frequency in Hertz
= Mechanical power in per-unit
= Electrical power in per-units

Figure 3 shows the flow chart of load shedding

algorithm. The algorithm will only initiate upon formation
of islanding. When the value of ROCOF greater than the
ROCOFMAX is detected, then the decision is to shed some of
the load. Power imbalance/disturbance magnitude is
calculated when the amount of load exceeds the DGs
capacity. To prevent further drops in frequency, the amount
of load require by disturbance magnitude must be shed.

Figure 1: Principle Operation of Load Shedding

489

2014 IEEE 8th International Power Engineering and Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah,
Malaysia. 24-25 March 2014

START

TABLE 1
LOAD RANKING TABLE

Measure the DG frequency

Is distribution
network islanded?

Load
Name

Measure the DG frequency

Capture the absolute

value of ROCOF
(ROCOF>ROCOFMAX)

Calculate total power imbalance

by using swing equation

Base Load

Peak Load

Load
Category

P
(MW)

Q
(MVAR)

P
(MW)

Q
(MVAR)

L1

0.0585

0.0372

0.0762

0.0471

L2

0.0798

0.0495

0.1038

0.0642

L3

0.1680

0.1042

0.2184

0.1353

L4

0.1878

0.1098

0.2204

0.1458

L5

0.3918

0.2427

0.4524

0.2756

L6

0.5178

0.3210

0.6058

0.3673

Vital

L7

0.5598

0.3468

0.6427

0.4009

Vital

1.9635

1.2112

2.3197

1.4362

NO

YES

Imbalance power greater

than DG rated

Load
Ranked

TOTAL

Non
Vital
Non
Vital
Non
Vital
Semi
Vital
Semi
Vital

NO

III.

RESULTS AND DISCUSSIONS

YES

A. Case Studies
The cases are tested on the test system as in Figure 4.
During grid connected, a total power of 1.78 MW is
dispatched by the generator with the remaining power is
dispatched by the grid. During islanding, the load in the
island is only supplied by a mini hydro generator (DG). In
an attempt to test UFLS technique, an intentional islanding
is simulated with the opening of incoming grid substation
breaker at time t=10.5 seconds (s). Load L4 is temporarily
disconnected from the system and only reconnected at
simulation time t=25.5s. Three cases are evaluated: base
load, peak load and overload case.

Shedding the load intelligently

(According to the load ranking table in Table 1)

END

Figure 3: Flow chart of UFLS on Response-Based Method

B. Test Network
a) Test system
The test system as shown in Figure 4 was modeled
using PSCAD software. The system consists 7 buses, 8
lumped load with 2 MVA mini hydro generation operated
at 3.3kV. The DG is connected via generator transformer
which step up the voltage level to 11.0kV. The distribution
network is connected to the main grid via two parallel step
up transformer (33kV/11kV) which is rated at 20MVA. This
test system has 17 remotes circuit breaker (RCB). That RCB
consists of a DG breaker, an islanded breaker, 2 breakers for
incoming grid substation, 8 breakers for load feeder and 5
breakers for additional breaker. The network was designed
using standard model of synchronous generator, exciter and
governor provided in PSCAD library.

a) Base load scenario

Simulation results for this case are shown in Figure 5, 6
and 7.
For base load scenario, after islanding the system
frequency rises to 50.5 Hz and then slowly recovers to
50.0Hz. The rise of frequency shows that the amount of
power generation is higher than total load. As shown in
Figure 5, without load shedding scheme, when L4 is
connected to the system at t=25.5s, the frequency decline
until 47.31Hz and fail to recover to its nominal frequency. It
was discussed in [10] that the generator will damage and
suddenly breakdown if the frequency continuously dropping
below 47.5 Hz. Therefore, certain amounts of load need to
be shed to improve the frequency response. By using the
LSCM, the disturbance magnitude/power imbalance is
identified.

b) Load
Based on test network, there are 7 loads (L1-L7)
and 1 addload (more explanation in section III) connected to
the network. The load is categorized into non-vital, semivital and vital as shown in Table 1. Non-vital load category
is less important load compared to semi-vital load and vital
load is the most important load in the system. The load will
be shed according to priority of load where less important
load will be shed first.
3

490

2014 IEEE 8th International Power Engineering and Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah,
Malaysia. 24-25 March 2014

Figure 4: Test System

After considering the spinning reserve, there is only one

load need to be shed i.e L1. This is corrresponding to load
ranking table as shown in Table 1. As shown in Table 2, the
power imbalance for this case is 0.0871M
MW and the total
amount has been shed by LSCM is 0.0585MW. As a result
of load shedding, the frequency drop too 48.57 Hz (refer
Figure 6) and recover to its nominal frequeency.

Figure 7: The amount of pow

wer imbalance for base load

b) Peak load scenario

In this case, not only loadd L4 is disconnected from the
system prior to islanding, but also load L5. To observe the
impact of load reconnection and the performance of load
shedding scheme, these two load are reconnected to the
system at t=25.5s. Simulation results
r
for this case are shown
in Figure 8, 9 and 10.
Without any load sheddding algorithm, the load
reconnection lead to a rapid decrement
d
in frequency which
eventually reach 18Hz. Obviiously, the frequency fail to
return to its nominal. In this caase, the under-frequency relay
of generator protection has already triggered at the moment
the frequency reach 47.5Hz. However,
H
when load shedding
scheme is applied, three loadss are shed i.e L1-L3. In this
case, the amount of load to be
b shed is determined by the
LSCM with regards to power imbalance as shown in Table
2. The power imbalance is 0.42275MW (refer Figure 10) and
the total amount of load sheddding by LSCM is 0.3984MW.
As a result, the frequency dropps to 48.24Hz and manage to
return at 50Hz (as shown in Figgure 9).

s
for base load
Figure 5 : System frequency response without load shedding

Figure 6 : System frequency response with load shhedding for base load

491

2014 IEEE 8th International Power Engineering and Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah,
Malaysia. 24-25 March 2014

power imbalance after Addloaad connection is 0.2613MW

while the total amount of load shedding
s
is 0.2478MW.

Figure 8 : System frequency response without load shedding

s
for peak load

Figure 11 : System frequency responsse with load shedding for increasing

loaad

Figure 9 : System frequency response with load shhedding for peak load

Figure 12: The amount of imballance power for increasing load

IV.

This paper presents an Undder-Frequency Load Shedding

(UFLS) technique for islandeed distribution network. This
technique focused on respponse based method. The
disturbance magnitude or poweer imbalance for the response
based is determined based on swing equation with the Rate
of Change of Frequency (ROCOF)
(
information. The
algorithm of UFLS technique has been developed by using
the PSCAD simulation tool. Thhe effectiveness of the scheme
has been investigated from
m a simulation study on
distribution system. Islanding is simulated when a part of
mmediately islanded from the
the distribution system is im
grid system. Three load condittions i.e base load, peak load
and overload case has been evaaluated. The simulation results
have shown that the technique has successfully shed the load
according to power imbalance and bring the frequency back
to its nominal (50Hz) even inn a different load condition.
The technique immediately performed load shedding
following disturbance and the load has been shed based on
load priority. The less importannt load has been shed first. As
a result, this technique was able to maintain the power
quality of the island thus allowing the island to persist.

Figure 10: The amount of imbalance power for peak load

TABLE 2
OPTIMAL LOAD SHEDDING BY LSCM
L
Load Case

Power Imbalance (MW)

Load Shed (MW)

Base Load
Peak Load
Overload

0.0871
0.4275
0.2613

0.0585
0.3984
0.2478

CON
NCLUSION

c) Overload scenario
In this case, an overload scenario is simulated
s
for base
case study. In contrast to base case, an
a additional load
referred as Addload (P=0.300MW and Q=
=0.1857MVAR) is
connected to the network at time t=45.55s. Obviously, the
excess of load would lead to frequency instability, unless
load shedding is performed. LSCM perform
m load shedding if
the estimated value of power imbalance exxceeds the unused
capacity of generator. In this case, two looad i.e L2 and L3
are shed. As a result, the frequency respponse manages to
restore. It is observed from the results thatt, the estimation of

ACKNOWLEEDGEMENT
This work was supportted by the University of
Technology MARA, Malaysiia under RIF grant project
(Grant Code: 600-RMI/DANA 5/3/RIF(366/2012)).

492

2014 IEEE 8th International Power Engineering and Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah,
Malaysia. 24-25 March 2014

REFERENCES
[1]
[2]

[3]

[4]
[5]
[6]
[7]

[8]
[9]

[10]

D. J. Pepermans G., Haeseldonckx D., Dhaeseleer., W.,

Belmans R., "Distributed Generation:Definition, Benefits and
Issues," University of Leuven Energy InstituteAugust 2003.
H. Seyedi and M. Sanaye-Pasand, "New centralised adaptive
load-shedding algorithms to mitigate power system blackouts,"
Generation, Transmission & Distribution, IET, vol. 3, pp. 99114, 2009.
T. Tomsic, G. Verbic, and F. Gubina, "Revision of the
underfrequency load-shedding scheme of the Slovenian power
system," Electric Power Systems Research, vol. 77, pp. 494-500,
2007.
J. G. Thompson and B. Fox, "Adaptive load shedding for
isolated power systems," Generation, Transmission and
Distribution, IEE Proceedings-, vol. 141, pp. 491-496, 1994.
C. Concordia, L. H. Fink, and G. Poullikkas, "Load shedding on
an isolated system," Power Systems, IEEE Transactions on, vol.
10, pp. 1467-1472, 1995.
C.-C. Huang and S.-J. Huang, "A time-based load shedding
protection for isolated power systems," Electric Power Systems
Research, vol. 52, pp. 161-169, 1999.
P. Mahat, C. Zhe, and B. Bak-Jensen, "Underfrequency Load
Shedding for an Islanded Distribution System With Distributed
Generators," Power Delivery, IEEE Transactions on, vol. 25,
pp. 911-918, 2010.
E. E. Aponte and J. K. Nelson, "Time optimal load shedding for
distributed power systems," Power Systems, IEEE Transactions
on, vol. 21, pp. 269-277, 2006.
M. Karimi, H. Mohamad, H. Mokhlis, and A. H. A. Bakar,
"Under-Frequency Load Shedding scheme for islanded
distribution network connected with mini hydro," International
Journal of Electrical Power & Energy Systems, vol. 42, pp. 127138, 2012.
A. A. M. Zin, H. M. Hafiz, and M. S. Aziz, "A review of underfrequency load shedding scheme on TNB system," in Power and
Energy Conference, 2004. PECon 2004. Proceedings. National,
2004, pp. 170-174.

493