scispace - formally typeset
Open AccessJournal ArticleDOI

Demand as Frequency Controlled Reserve

TLDR
In this paper, the authors investigated the use of the electricity demand as frequency controlled reserve (DFR) for future Danish power system, where a 50% wind penetration is outlined by the government for year 2025.
Abstract
Relying on generation side alone is deemed insufficient to fulfill the system balancing needs for future Danish power system, where a 50% wind penetration is outlined by the government for year 2025. This paper investigates use of the electricity demand as frequency controlled reserve (DFR), which has a high potential and can provide many advantages. Firstly, the background of the research is reviewed, including conventional power system reserves and the demand side potentials. Subsequently, the control logics and corresponding design considerations for the DFR technology have been developed and analyzed, based on which simulation models have been built using the DIgSILENT Power Factory. The simulation studies of different scenarios confirm that the DFR can provide reliable performance of frequency control. Furthermore, relevant issues regarding implementing DFR in reality have been discussed.

read more

Content maybe subject to copyright    Report

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from orbit.dtu.dk on: Aug 10, 2022
Demand as Frequency Controlled Reserve
Xu, Zhao; Østergaard, Jacob; Togeby, Mikael
Published in:
IEEE Transactions on Power Systems
Link to article, DOI:
10.1109/TPWRS.2010.2080293
Publication date:
2011
Link back to DTU Orbit
Citation (APA):
Xu, Z., Østergaard, J., & Togeby, M. (2011). Demand as Frequency Controlled Reserve. IEEE Transactions on
Power Systems, 26(3), 1062-1071. https://doi.org/10.1109/TPWRS.2010.2080293

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON POWER SYSTEMS 1
Demand as Frequency Controlled Reserve
Zhao Xu, Member, IEEE, Jacob Østergaard, Senior Member, IEEE, and Mikael Togeby
Abstract—Relying on generation side alone is deemed insuffi-
cient to fulfill the system balancing needs for future Danish power
system, where a 50% wind penetration is outlined by the govern-
ment for year 2025. This paper investigates use of the electricity de-
mand as frequency controlled reserve (DFR), which has a high po-
tential and can provide many advantages. Firstly, the background
of the research is reviewed, including conventional power system
reserves and the demand side potentials. Subsequently, the control
logics and corresponding design considerations for the DFR tech-
nology have been developed and analyzed, based on which simula-
tion models have been built using the DIgSILENT Power Factory.
The simulation studies of different scenarios confirm that the DFR
can provide reliable performance of frequency control. Further-
more, relevant issues regarding implementing DFR in reality have
been discussed.
Index Terms—Demand side, frequency control, power system
modeling and simulation, wind power.
I. INTRODUCTION
M
AINTAINING the balance between the power gener-
ation and the demand, i.e., the system frequency, is a
key issue in power system operation. Today, generation side re-
sources including extra capacities from large generators and in-
terconnection lines are primarily used for such purpose. Discon-
nection of demand, i.e., load shedding, is used as the last resort
in emergency situation and whole areas are shed if the frequency
drops below a very low limit, for example, 49.50 Hz or lower in
the Nordic interconnected system [1].
Theoretically, the generation and demand can contribute
equally to the frequency control as reserves. However, the
demand has so far been largely underutilized. This is mostly
because the current control paradigm was designed to procure
services from several large power plants, with prejudices
against demand side resources. The complexity of fulfilling
the real-time monitoring requirement for many distributed,
small sized loads, is considered the major obstacle to utilize the
demand as reserves. Customer comforts and wear and tear are
also concerned for end user appliances. Besides, the business
Manuscript received October 16, 2009; revised October 21, 2009, February
25, 2010, June 03, 2010, and August 07, 2010; accepted September 13, 2010.
This work was supported by the Danish Public Service Obligation (PSO) re-
search funding program, Project “Demand as Frequency Controlled Reserve”,
Grant no. 2005-2-6380. Paper no. TPWRS-00818-2009.
Z. Xu is with the Department of Electrical Engineering, Hong Kong Poly-
technic University, Hunghom, Kowloon, Hong Kong (e-mail: eezhaoxu@polyu.
edu.hk).
J. Østergaard is with the Centre for Electric Technology, Department of Elec-
trical Engineering, Technical University of Denmark, DK-2800 Lyngby, Den-
mark (e-mail: joe@elektro.dtu.dk).
M. Togeby is with the Ea Energianalyse A/S, Frederiksholm Kanal 1, 1220
Copenhagen, Denmark (e-mail: mt@eaea.dk).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRS.2010.2080293
model is another critical concern when dealing with many
small loads.
In fact, many electricity demands, though with small ca-
pacities, can be turned on and off as frequently, and rapidly
as needed, rendering perfect capability to be used as fast re-
serves. By installing with a frequency sensor and appropriate
control intelligence, those loads can respond autonomously
to frequency variation and provide fast reserve to the system.
Household appliances including electric heating, refrigerators,
freezers, and water heaters are ideal candidates due to their con-
siderable volume and the possibility of instantaneous switching
off. Accurate monitoring of such resources (e.g., with online
meters) may be impractical, but estimation is possible, which
is consistent with how the demand is treated in the market
dispatching today. Compared with traditional reserves, the
DFR can deliver the same service plus additional benefits, e.g.,
faster response speed, potentially lower costs, well-dispersed
in distribution network and pollution free, etc. [2], [3]. This is
particularly important for countries like Denmark, where lack
of balancing resources is foreseen as the major challenge to
achieve 50% wind penetration by 2025 [4].
Efforts have been made in the field of using demand as re-
serves. In the past, utilities have long been operating load man-
agement programs [6]. In [7], a market-based demand man-
agement program using low frequency relay to control indus-
trial loads is reported. A similar program is implemented in the
New Zealand power system [8]. In Finland, 1000-MW demands
from wood processing, chemical, and metal industry are used
as frequency controlled as well as manual reserves [9]. These
programs have, however, mainly focused on large size industry
loads. A pilot project using the
ComfortChoice technology for
controlling air conditioners to provide reserve was carried out by
the Long Island Power Authority in 2003 [10]. Due to the com-
munication (two-way paging) system introduced, the reserve is
activated relatively slow in about 90 s.
The Pacific Northwest National Laboratory (PNNL) has sug-
gested that individual household appliances suitable for tempo-
rary disconnection can provide fast reserve within seconds, e.g.,
refrigerators and air conditioners [11], [12]. Similar suggestions
have been made in the U.K. [13]–[15]. The idea was presented
as early as in 1979 [16]. Because the implementation requires
little electronics at low costs [13], these ideas become more at-
tractive recently.
This paper investigates use of electricity demands as a
new measure for fast reserves including frequency controlled
disturbance and normal ones. Section II reviews the status
of active power reserves and frequency quality in the Nordic
power system. In Section III, different DFR control logics
and considerations have been made to address concerns of
both power system and customers. Section IV develops the
simulation models of DFR demands using DIgSILENT Power
0885-8950/$26.00 © 2010 IEEE

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
2 IEEE TRANSACTIONS ON POWER SYSTEMS
Fig. 1. Available frequency controlled normal and disturbance reserves in the
Nordic power system.
Factory, based on which various scenarios have been simulated
in Section V using practical power system models that demon-
strates the DFR can perform as well as the generation side
reserves. Section V outlines major conclusions as well as future
scope of our research work. Furthermore, issues concerning
monitoring method and market models for DFR have also been
preliminarily discussed, yet to be continued in the future.
II. B
ACKGROUND AND
PROSPECTIVE
A. Status of Frequency Control in the Nordic Power System
The Nordic system has been investigated to identify specific
reserves that best suit the demand side. Other systems like the
UCTE of mainland EU will be studied in the future. In the
Nordic system that covers Eastern Denmark, Norway, Finland,
and Sweden, reserves are classified as frequency controlled
normal and disturbance reserves, and fast and slow active
disturbance reserves that mainly differ in the activation time
and frequencies [1]. The market for these reserves is introduced
in [17].
Using demand as disturbance reserve is focused, and the op-
tion of using demand as normal reserve is also studied. To-
tally 1000-MW disturbance reserve must be procured by the
Nordic TSOs, and in events of contingencies, the reserve must
be activated automatically and linearly between 49.5 to 49.9 Hz
[5]. The normal reserve deals with small frequency deviations
in normal condition. It consists of up and down regulation for
frequency variations between 49.9–50.1 Hz, and is automati-
cally activated within 2–3 min. The volumes of required fre-
quency controlled reserves are presented in Fig. 1, where the
linear activation of the disturbance reserve is also shown. The
fast and slow reserves are manually activated within or after 15
min, respectively, in order to replace the normal and disturbance
reserves. All reserves are mainly provided by extra capacities
from generators or HVDC lines, which are costly. E.g., the cost
of 50-MW reservation on the Kontek HVDC between Eastern
Denmark and Germany is estimated 125 000
/MW per year
based on absolute differences in spot prices from November
2005 to May 2010, and reserved capacity (for frequency reserve)
between the two areas.
From September 2009 at limited amount of frequency
controlled disturbance reserves (in average 8 MW) has been
purchased via a public auction. A reservation price is paid
for blocks of 4 h. The average reservation price (September
Fig. 2. Probability of frequencies over and under the normal range (49.9–50.1
Hz). Data are provided by Swedish TSO Svenska Kraftnät, recorded from Au-
gust 1995 to December 2009, at 1-min sample interval.
Fig. 3. Probability of frequencies under 49.90 Hz within the minutes of the
hour. (The data of
>
1
:
5 billion
samples are measured per 0.02 s for a year long
period from April 2005 to March 2006 using phase measurement units installed
in Eastern Danish grid [22], [23], and processed using DTU high performance
computing facility [24].)
16, 2009 to August 1, 2010, Denmark East) has been 250 000
/MW per year.
Due to the frequency dependency, the total demand of the
Nordic system reduces roughly by 200 MW when the frequency
drops from 50.0 to 49.5 Hz [1]. This is already considered into
the reserve dispatching. Therefore, using the demands as fast
reserves can be understood as to enhance their frequency re-
sponse by deliberately switching on and off the non-essential
ones. Well-designed, this can increase the stability of the elec-
tricity system.
The Nordic system has 49-GW hydropower in Norway
Sweden and Finland, corresponding to 52% of its installed
capacity, which is superior for frequency control [39]. Fig. 2
shows the probability of abnormal frequency (
or
) has been increasing throughout the years. The
probabilities since 2002 are above 0.8% which is more than
a factor 3 higher than the goal 0.228% [18]. In 2003, low
supplies of power with associated high prices can explain the
peak of abnormal frequency in Fig. 2 [19]. Another reason for
increased frequency excursions could be the reserve require-
ment was defined over 20 years ago and may not be up to date.
Furthermore, some hydro plants are not originally designed to
provide continuous frequency regulations [38].
Fig. 3 plots the probability of under-frequency versus the
minute in an hour, where the probability appears much higher in
the first 15 min of an hour compared to the rest. This can be well
explained as the Nordic spot market is dispatched hourly, and
consequently often intra-hour imbalances due to hourly shifts

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
XU et al.: DEMAND AS FREQUENCY CONTROLLED RESERVE 3
TABLE I
P
OTENTIALS FOR
HOUSEHOLD
ELECTRICITY
DEMAND AS
FREQUENCY CONTROLLED
RESERVE
* [25]
exist. Such a market pattern is also verified by analyzing fre-
quency data of UCTE and WECC systems, though relatively
smaller probability of low frequency is observed due to a larger
system size. The frequency analyses reveal that, besides failure
events like generator outages, other factors like the market can
cause abnormal frequencies as well. Notably, the wind penetra-
tion in the Nordic system is still marginal
. As Nordic
countries, especially Denmark, are developing more wind en-
ergy, power balancing control will become critical in system
operation. New measures such as the DFR technology should
be applied in this context [4], [20], [21].
B. Demand Side Prospective
Many demands that can be disconnected quickly and shortly
with limited disturbances to the appliances and customers com-
forts are suitable for DFR applications. The thermostatically
controlled loads are especially ideal as they can act as energy
storage. E.g., a few minutes of disconnection of electrical
heating or cooling appliances will reduce or increase the tem-
perature, but not to the extent to annoy the customers or harm
the stored content. The average consumption for an individual
appliance is limited, e.g., average consumption varies from
22 W, for a high efficiency refrigerator, to 400 W for water
heater in a four-person family. Electric space heating can be in
the order of 1 to 3 kW per household. Nevertheless, the total
compatible household demands constitute a large consumption
according to our survey in Eastern Denmark in Table I, where
218 MW of such demands is much larger than the required
78-MW disturbance reserve [25]. The potential in the overall
Nordic system can be much higher due to the large use of
electric heating in Norway, Sweden, and Finland.
Some compatible demands have constant consumption
during all hours of the year, when a large number of refrigera-
tors, freezers, or water heaters are considered. The variation due
to user interaction is limited and the seasonal variation is little.
Others like electric space heating have significant seasonal
variation, but can be reliably predicted when aggregated. For
appliances like washing machines and tumble driers, the daily
variation is high, typically with no demand at night. For electric
heating, disconnection and reconnection can happen instanta-
neously. For compressor type of demand, it is impossible to
reconnect right after a disconnection. The typical compressor
system in refrigerators and freezers are designed so they need
a minimum 5-min resting period before next reconnection.
Besides household appliances, compatible loads are found
in industrial sectors as well, e.g., electric ovens, ventilation
systems, centrifuges, and aerating at waste water treatment
plants. Especially, uninterruptible power supply (UPS) with
battery storage and control system can implement the frequency
reserve easily. It is estimated that in Danish companies 96-MW
demand can be interrupted with the shortest notice of 5 s [21].
Many modern appliances are equipped with advanced elec-
tronic control, potentially useful when implementing frequency
control. In some modern washing machines and tumble driers,
possibilities of stopping and delaying the process already exist.
In these cases, the cost of integrating the frequency control will
be limited. In other cases, like water heaters, a mechanical ther-
mostat is normally used to control the temperature and higher
investment cost for frequency control can be foreseen.
Traditional reserve can be very costly, e.g., the primary and
disturbance reserves in West and East Denmark are 22 and 8
k
/MW/year, respectively, in 2007 [26]. With an estimated
production cost of the frequency control circuit of 20
[12],
the cost of reserve from DFR is roughly 3.2 k
/MW/year for
the 1-kW appliance, 1.6 k
/MW/year for the 2-kW appliance,
based on eight-year lifetime and an annual interest rate of 6%.
Here the annuity values are calculated by
(1)
where
is annuity, is the initial cost, is the discount rate,
and
is payment period in years.
III. DFR C
ONTROL LOGICS
The DFR technology is to flexibly turn on or off electricity
loads in response to frequency variations to provide normal or
disturbance reserves. This can be implemented as an external
control box or integrated within the electric appliances. The con-
trol should be designed considering the requirements of power
systems, and the special features of the appliances, etc. Two
types of control logics to switch on/off demands directly or indi-
rectly are developed. Fig. 4 shows the control scheme with three
stages.
1) The scheme starts with measuring the frequency at a time
step, e.g., 0.02 s. A moving average filter of a length, e.g.,
0.5 s, can be used to reduce measurement noise. This will
be implemented with a sliding window algorithm for con-
tinuous measurement.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
4 IEEE TRANSACTIONS ON POWER SYSTEMS
Fig. 4. Overview of the DFR control scheme.
Fig. 5. Illustration of DFR Type I control and critical parameters.
2) Disconnection with delay will be activated based on the
comparison of the measurement and the disconnection cri-
terion. Otherwise, the scheme continues to the next cycle.
3) After disconnection, the scheme continues to measure and
compare the frequency with the reconnection criteria. If
satisfied, the appliances will be reconnected back to the
grid after a certain delay.
A. DFR Control Logic Type I
The Type I disconnects and reconnects electric appliances,
e.g., a water heater to the grid when system frequency falls and
recovers, respectively [2]. The reconnection frequency set point
should be equal to or higher than the disconnection set point
, to give a hysteresis that can prevent oscillatory behav-
iors and consequently excessive wear outs to the appliances [2]
(Fig. 5). Time delays
and are introduced in the discon-
nection and reconnection to deal with the measurement noise
and other considerations.
The disconnection set point can be designed in accordance
with the specifications of the targeted reserve, e.g., 49.90 and
49.95 Hz for frequency controlled disturbance and normal re-
serves, respectively, in the Nordic power system. To achieve
proportional activation and prevent over-activation of DFR,
randomized set points within, e.g.,
can be
applied to a large group of appliances. For the thermostatically
controlled loads, two variant designs are developed as in Table II.
When reconnected, Type
always starts “on”, while Type
starts according to the appliance temperature, e.g., if the temper-
ature is above (below) the lower limit, the heaters will starts with
“off” (on). They can give different impacts to the power system.
TABLE II
V
ARIANT
DESIGNS FOR
DFR TYPE I C
ONTROL
B. DFR Control Logic Type II
The Type II control is specially designed for thermostatically
controlled loads to indirectly switch the loads by adjusting the
temperature set points [13]. For example, a linear relationship
between the temperature set points and system frequency can
be used for electric heating loads, i.e.,
(2)
(3)
where
and are nominal high and low temper-
ature set points,
is the nominal frequency which is 50 Hz for
the Nordic system,
is the measured frequency, and is
the coefficient of frequency change in
.
Consider a large number of electric heaters in normal oper-
ation; they can reach the equilibrium status with their temper-
atures well distributed within the upper and lower limits. The
number of on and off units will become relatively stable, so
as the total consumption power of “on” units. If the frequency
varies by
and if , the power
decrease or increase due to Type II control can be estimated by
(4) and (5), respectively:
(4)
(5)
where
is total installation power of heaters, represents the
percentage of on heaters. If the frequency falls or rises, the con-
trol will start disconnecting or reconnecting appliances close to
the end of their on or off cycle. Due to the internal thermal con-
trol, the disconnected appliances will be reconnected sooner or
later regardless of the frequency. Therefore, Type II control can
be less disturbing to the end-user as it only changes the appli-
ance operation cycles for providing reserves. Because not all
heaters are always on, the reserve from Type II has to be re-
alized by aggregating many of such appliances, of which only
a small portion will be affected in most low or high frequency
events, with limited disturbance to customer comforts.
Proper design of
should consider the requirements from
power systems and appliances, etc. Moreover, different
values can be used for different frequency deviations. E.g.,
a smaller coefficient is used for small deviations between
under normal operation, and a bigger co-
efficient is used for large deviations
to ensure as
many DFR units as possible to be activated. In this paper, only
single coefficient is considered, and more advanced design will
be investigated in the future. Similar control can be applied
for the cooling appliances like refrigerators or freezers, though
slight changes may be necessary due to, e.g., only the lower
temperature set point can be varied for industrial freezers.

Citations
More filters
Proceedings ArticleDOI

Foundations and Challenges of Low-Inertia Systems (Invited Paper)

TL;DR: The challenges of such low-inertia power systems are reviewed, the solutions that have been put forward thus far are surveyed, and the topics of power system stability, modeling, and control are touched upon.
Journal ArticleDOI

Overview of energy storage systems in distribution networks: Placement, sizing, operation, and power quality

TL;DR: This study identifies future research opportunities in relation to challenges for optimal ESS placement planning, development and implementation issues, optimisation techniques, social impacts, and energy security.
Journal ArticleDOI

Load management: Model-based control of aggregate power for populations of thermostatically controlled loads

TL;DR: In this paper, the authors developed a mathematical model aimed at the design of a model-based feedback control strategy, which analytically characterises the aggregate power response of a population of ACs to a simultaneous step change in temperature set points.
Journal ArticleDOI

Demand response in liberalized electricity markets: Analysis of aggregated load participation in the German balancing mechanism

TL;DR: In this article, an analysis of the German balancing mechanism illustrates that DR is undermined by three mechanism design aspects: minimum bidding volume, minimum bid duration and binding up and down bids.
Journal ArticleDOI

Modeling and Analysis of a Variable Speed Heat Pump for Frequency Regulation Through Direct Load Control

TL;DR: In this article, a dynamic model of a variable speed heat pump (VSHP) in a commercial building that responds to direct load control (DLC) signals, updated every 4 s, for the improvement of grid frequency regulation (GFR) is presented.
References
More filters
Book

Power System Stability and Control

P. Kundur
TL;DR: In this article, the authors present a model for the power system stability problem in modern power systems based on Synchronous Machine Theory and Modelling, and a model representation of the synchronous machine representation in stability studies.
Journal ArticleDOI

Stabilization of Grid Frequency Through Dynamic Demand Control

TL;DR: In this paper, the authors investigated whether a degree of built-in frequency stability could be provided by incorporating dynamic demand control into certain consumer appliances, such as refrigerators, which would monitor system frequency and switch the appliance on or off accordingly, striking a compromise between the needs of the appliance and the grid.
Journal ArticleDOI

Decentralized Demand-Side Contribution to Primary Frequency Control

TL;DR: In this paper, the authors argue that two-way communication is not essential and that the demand can respond to the frequency error in a manner similar to the generators, and that, using this approach, the demand side can make a significant and reliable contribution to primary frequency response while preserving the benefits that consumers derive from their supply of electric energy.
Journal ArticleDOI

A state-queueing model of thermostatically controlled appliances

TL;DR: In this article, the authors developed a state-queueing model to analyze the price response of aggregated loads consisting of thermostatically controlled appliances (TCAs), and they showed that TCA setpoint changes in response to the market price will result in a redistribution of TCAs in on/off states and therefore change the probabilities for a unit to reside in each state.
Journal ArticleDOI

Modeling uncertainties in aggregated thermostatically controlled loads using a State queueing model

TL;DR: In this article, the modeling of uncertainties in aggregated thermostatically controlled loads using a state queueing (SQ) model is discussed, and the impacts of considering load diversity in the SQ model while simulating TCA setpoint response are also studied.
Related Papers (5)