Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 88 (2016) 436 – 442

CUE2015-Applied Energy Symposium and Summit 2015: Low carbon cities and urban
energy systems

Virtual Energy Storage System for Smart Grids

Meng Chenga*, Saif Sabah Samia, Jianzhong Wua
a
Cardiff School of Engineering, Cardiff University, Cardiff, CF24 3AA, UK

Abstract

This paper forms a Virtual Energy Storage System (VESS) and validates that VESS is a cost-effective way
to provide the function of energy storage through the utilization of the present network assets represented by flexible
demand. As a solution to convert to low carbon cities, a VESS is firstly modelled to store and release energy in
response to regulation signals by coordinating the demand response (DR) from domestic refrigerators in London and
the conventional flywheel energy storage systems (FESS). The coordination of DR and FESS mitigates the
uncertainties of DR and reduces the capacity of costly FESS. The VESS is applied to provide ancillary services to the
power system and contributes to the reduction of carbon emission through the replacement of spinning reserve
capacity of fossil fuel generators. Case studies were carried out to validate and quantify the capability of the VESS to
vary the stored energy in response to grid frequency. Economic benefits of using VESS for frequency response
services were firstly estimated and a potential saving of £91m-£103m is expected.

©
© 2016
2015The TheAuthors. Published
Authors. by Elsevier
Published Ltd. This
by Elsevier Ltd.is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Selection and/or peer-review under responsibility of CUE
Peer-review under responsibility of the organizing committee of CUE 2015

Keywords: Virtual energy storage system, Low carbon cities, Smart grids

Nomenclature
Abbreviation
DR Demand Response
ESS Energy Storage system
FESS Flywheel Energy Storage System
RES Renewable Energy Source
VESS Virtual Energy Storage System

* Meng Cheng. Tel.: +44 (0)2920870422.

E-mail address: chengm2@cardiff.ac.uk

1876-6102 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of CUE 2015
doi:10.1016/j.egypro.2016.06.021
 Meng Cheng et al. / Energy Procedia 88 (2016) 436 – 442 437

1. Introduction

Cities are rapidly integrating smart grid technologies to move towards an energy efficient future with
lower carbon emissions. The increasing integration of renewable energy sources (RES), such as the
photovoltaic and the wind, causes uncertainties in electricity supply. It is therefore more challengeable to
meet the power system demand. More reserve is required from partly-loaded fossil-fuel generators which
are usually costly and exacerbate the carbon emissions.
In addition, the integration of RES through power electronics reduces the system inertia. Frequency
indicates the balance between generation and demand. A low inertia power system will encounter more
severe frequency stability issues in cases of sudden changes in supply or demand [1]. Therefore, faster
response to frequency changes is required.
Energy Storage System (ESS) is one solution to facilitate the integration of RES by immediately
varying the stored energy. In terms of the functions of ESS, ESS is classified as high power rating for
power quality applications and high energy rating for energy management applications. In terms of the
forms of ESS, ESS is classified as mechanical, chemical, electrical and thermal categories [2]-[3].
However, ESS remains to be an expensive technology although in recent there is declination in the
cost. For instance, the cost of installing a 20MW/10MWh Flywheel Energy Storage Systems (FESS) is
approx. £25m-£28m [4].
Aggregated Demand Response (DR) can act as virtual energy storage because DR can provide
functions similar to the energy storage by intelligently managing the power and energy consumption of
loads. By utilizing the existing network assets, DR can be deployed at scale with lower cost. A cost of
£1.97/household/year of using domestic refrigerators and freezers for frequency response is shown in [5]
based on the 2013 GB electricity market rates. Considering the availability of refrigerators to provide
frequency response [6], it is estimated that 20 MW of response requires approx. 1.5m refrigerators.
Therefore, the total cost of 20 MW of DR is 3m. This is far smaller than the cost of FESS (approx. £25m-
£28m [4]) to also provide the 20 MW of response. It is estimated in [4] that DR has the potential to reduce
the ESS market size by 50% in 2030.
However, the challenges of DR include the uncertainty of the response and the consequent reduction in
the diversity amongst loads [6]. Simultaneous connection of loads may occur in several minutes after the
provision of response to a frequency drop. This will worsen the system stability.
In this paper, a VESS is formed as a single entity to provide functions of energy storage. The VESS
utilizes the capacity of DR to reduce the capacity of costly ESS whilst reducing the impact of DR’s
uncertainties. A VESS consisting of DR from domestic refrigerators in London and of FESS is modelled
and controlled. The proposed control of VESS maintains the load diversity and the primary functions of
cold storage of refrigerators, and reduces the number of charging and discharging of each FESS. Case
studies were carried out to quantify the capability of VESS for frequency response services and to
compare the capability of VESS with FESS. The potential economic benefit is also estimated.

2. Virtual Energy Storage System

2.1. Concept
A VESS aggregates a cluster of flexible loads, ESS and can also extracts energy from Distributed
Generators via smart grid technologies. Through the coordination of each unit, a VESS acts as a single
high capacity ESS with lower capital costs.
A VESS allows the small-capacity ESS, flexible loads and distributed energy resources to get access to
the wholesale market and to provide transmission level services to the power system. Because VESS
forms a synthetic ESS at the transmission/distribution level through the aggregation, the high power and
438 Meng Cheng et al. / Energy Procedia 88 (2016) 436 – 442

high energy ratings of the VESS makes it suitable for a wide spectrum of applications including the
maintenance of power quality, power system stability and protection, energy management and the
mitigation of network reinforcements.
Compared to the Virtual Power Plant concept that aggregates DGs to act as a single power plant, VESS
aims to store surplus electricity or release electricity according to system needs.

2.2. VESS model and control

A VESS model is formed to provide a required amount of frequency response (similar to an ESS) to the
power system in order to participate in the GB frequency response market as an aggregator. DR of
domestic refrigerators in a city is implemented and conventional FESS is used to compensate for the
uncertainties caused by DR. Other units with similar characteristics can be added further to increase the
total energy storage capacity.

2.2.1. Model of FESS

FESS is a mechanical ESS with high power density. An FESS is formed as an electrical machine
connecting to the grid through back-to-back converters. The electromechanical equation of the electrical
machine is:

dZ Pout
J Tin  (1)
dt Z
where J (Kg.m2) is the flywheel inertia, ω (rad/s) is the rotating speed, Tin (Nm) is the input torque of
the flywheel, Pout (Watts) is the output power that is controlled by converters (Pelec).
A first-order lag is used to simplify the control of the power converter on given reference power Pref
(Watts). Tdelay is the time constant of the power converter control loop

§ 1 ·
Pout Pref ¨ ¸ (2)
¨ 1  sT ¸
© delay ¹

A simplified model of an FESS was developed as shown in Figure 1. It has been validated with a
detailed model which includes all the elements and control. The simplified model provided accurate
results with significant reduction in computational time. It facilitates the system level studies considering
large numbers of small flywheels.
Flywheels
Internal Control T Pmech
Pref 1 1 Pelec
Tdelay 1 ω
s
Js

Figure 1: A simplified model of a FESS

2.2.2. Model of domestic refrigerators

In order not to undermine the cold storage function of each refrigerator, a thermodynamic model of
refrigerators was developed as illustrated in [6].
The internal temperature (T) of a refrigerator is modelled and dynamically compared with the
temperature set-points Tlow and Thigh. When a refrigerator is ON, it is equivalent to charging an energy
 Meng Cheng et al. / Energy Procedia 88 (2016) 436 – 442 439

storage which decreases T. An OFF-state refrigerator is considered as a discharging process that increases
T. In a refrigerator, temperature controls the charging and discharging process.

2.2.3. Distributed control in a VESS

A control algorithm is developed for VESS to provide certain amount of response to regulation signals.
In this paper, grid frequency (f) is used as the regulation signal.
A general local controller for refrigerators and FESS is developed as shown in Figure 2. The control
measures f constantly. For an FESS, the output is the change of power output. For a refrigerator, the
output is the change of On/Off state and hence the power consumption.
Each unit in the VESS is assigned a pair of frequency set-points, FON and FOFF. The range of FON is 50
– 50.5 Hz and the range of FOFF is 49.5 – 50 Hz which is consistent with the steady-state limits of grid
frequency in the UK.
If f is higher than FON of a unit, the unit will start charging/switch on as a result of the frequency rise.
If f is higher than 50 Hz but lower than FON, the unit will standby.
If f is lower than FOFF, the unit will start discharging/switch off as a result of the frequency drop. If f is
lower than 50 Hz but higher than FOFF, the unit will standby.
FON and FOFF vary linearly with the State of Charge (SoC) of each unit as shown in Figure 2. For
FESS, ω indicates SoC. For refrigerators, T indicates SoC. A higher T indicates a lower SoC and vice
versa. When f rises, the units will start charging from the one with the lowest SoC. When f drops, the
units will start discharging from the one with the highest SoC. Hence, the more f drops, the more power
will be generated from FESS and the more power consumption of refrigerators will be reduced. Vice
versa for f rises.
The final state of a unit is determined by a set of logic gates. The inherent control of each unit takes the
priority to make the decision. The temperature control of refrigerators and the charging control of FESS
limiting the maximum and minimum ω are the inherent control.
The control in Figure 2 is a local control on each unit, but is coordinated by assigning frequency set-
points based on SoC. This reduces the number of charging/discharging cycles and each unit has equal
opportunity to charge/discharge. The lifetime of units is hence prolonged. Specifically for refrigerators,
the control does not undermine the cold storage. The diversity amongst refrigerators is maintained.

Figure 2: Controller of VESS

440 Meng Cheng et al. / Energy Procedia 88 (2016) 436 – 442

2.2.4. Coordination of refrigerators and FESS in a VESS

As discussed in Section 1, the high cost of FESS limits the grid-scale deployment. The uncertainty of
DR makes it difficult to ensure the provision of certain amount of response. The coordination of FESS
and DR in a VESS aims to provide certain amount of frequency response at low costs.
For refrigerators, f in Figure 2 is grid frequency. For FESS, f is the output of a droop setting as
illustrated below.
The difference between the required frequency response of VESS (Preq) and the response of
refrigerators (Pr) is input to the droop setting. The droop parameter (R) is:

50 Hz  49.5Hz
R (3)
PFESS

where PFESS (MW) is the total capacity of FESS.

Namely, FESS responds to the mismatch between Preq and Pr in order to compensate for the uncertain
response of refrigerators.

3. Case Study – Smart city, London

There are 3,220,300 households in London in 2014 [7]. It is assumed that the refrigerator (0.1 kW) in
each household is equipped with the frequency controller in section 2.2.
The amount of frequency response from refrigerators is estimated considering the time of day [6]. A
maximum reduction in power consumption is 18.5% at 18:00 and a minimum reduction is 13.2% at 6:00.
Considering the number of refrigerators in London, a maximum power reduction of 60 MW and a
minimum power reduction of 40 MW is possible.
Therefore, VESS is promised to provide a linear frequency response of a maximum of 60 MW (Preq)
to the power system over a day when frequency drops outside the limit (49.5 Hz). For the periods that
refrigerators cannot provide 60 MW of response, FESS compensates for the maximum mismatch of 20
MW.
The following three case studies were undertaken on a GB power system model representing a low
inertia future power system as developed in [8]. System base was 20 GW.
ƒ Case1 assumes there is no VESS.
ƒ Case2 connects 1,200 FESS with the rating of 50 kW. This case uses conventional FESS to
provide the 60 MW of frequency response.
ƒ Case3 connects the VESS including refrigerators in London and 400 FESS with the rating of 50
kW. The FESSs provide 20 MW of response to the mismatch between Preq (60 MW) and Pr.
Simulations were carried out by applying a loss of generation of 1.8 GW in each of the three cases.

4. Results and Discussions

In Figures 3-4, the frequency drop is reduced with 60 MW of response from either FESS (case2) or
VESS (case3). As 60 MW is small in a 20 GW system, the improvements of frequency is not obvious. If
the capacity of VESS is higher, frequency drop will be more greatly reduced. The number of FESS in
case3 was only one third of that in case2, however, VESS provided similar amount of frequency response
as in case2. The reduced capacity of FESS in case3 will reduce the cost significantly compared to case2.
Figure 5 depicts that power consumption of refrigerators (left axis) reduced from 62 MW to 22 MW
following the frequency drop. After 15 min, power consumption of refrigerators recovers and exceeds the
initial 62 MW. FESS in the VESS increases the power output to reduce the effect of reconnections of
 Meng Cheng et al. / Energy Procedia 88 (2016) 436 – 442 441

refrigerators. The total frequency response from VESS is in linear relationship with frequency deviations
as indicated in Figures 3-4.
50.1
50
49.9

Grid Frequency (Hz)

49.9
49.8
49.7 49.7
49.6 49.5
49.5
49.4 No VESS (case1) 49.3
49.3 Time (s)
with FESS (case2) 49.1
49.2 with VESS (case3) 0 5 10 15
49.1
0 5 10 15 20 25 30
Time (min)

Figure 3: Variation of grid frequency

70
Frequency Response (MW)

60
50
40
30
Flywheels (case2)
20
10 VESS (case 3)
0
0 5 10 15 20 25 30
Time (min)
Figure 4 Frequency response from FESS (case2) and VESS (case3) 30 min after the generation loss

To provide frequency response of 60 MW, the total cost of FESS in case2 is £150m-£168m [4]. If the
VESS is used for 60 MW of frequency response, the cost of refrigerators (40-60 MW) is £9.66m [9] and
the cost of FESS (20 MW) is £50m-£56m. A potential saving of £91m-£103m is obtained.

100 25
refrigerator initial power
Power Output of Flywheels

20
Refrigerators Power
Consumption (MW)

80 consumption 62 MW
15
60 10
(MW)

5
40 0
-5
20 Refrigerators (case3)
-10
FESS (case3)
0 -15
0 5 10 15 20 25 30
Time (min)

Figure 5 Change in power consumption of refrigerators and power output of FESS in VESS (case3)

5. Conclusions
To facilitate the integration of RES and move towards a low carbon city, a VESS is formed by
coordinating the DR of domestic refrigerators and FESS to provide functions similar to conventional ESS
with higher capacity and lower cost. A model and control algorithm of the VESS was developed. The
control minimizes the charging/discharging cycles of each unit and hence prolongs the lifetime of each
unit. The control also maintains the primary function of loads and maintains the load diversity amongst
population. Case studies were undertaken to evaluate the capability of VESS for frequency response by
442 Meng Cheng et al. / Energy Procedia 88 (2016) 436 – 442

connecting to a simplified GB power system model. Simulation results showed that VESS provides
frequency response in a manner similar to FESS. Compared with the case that only uses FESS to provide
a 60 MW of frequency response, a potential saving of around £91m-£103m was obtained by employing
the VESS.

Acknowledgement
The work was supported in part by RESTORES project under the grant of EPSRC, by P2P-SmarTest
project under the grant of EU commission and by Higher Committee for Education Development in Iraq
(HCED).

References
[1] Zhao H, Wu Q, Hu S, Xu H, Rasmussen CN. Review of energy storage system for wind power
integration support. Appl Energy 2015;137:545–53. doi:10.1016/j.apenergy.2014.04.103.
[2] Chen H, Cong TN, Yang W, Tan C, Li Y, Ding Y. Progress in electrical energy storage system: A
critical review. Prog Nat Sci 2009;19:291–312. doi:http://dx.doi.org/10.1016/j.pnsc.2008.07.014.
[3] Luo X, Wang J, Dooner M, Clarke J. Overview of current development in electrical energy
storage technologies and the application potential in power system operation. Appl Energy
2014;137:511–36. doi:10.1016/j.apenergy.2014.09.081.
[4] Strbac G, Aunedi M, Pudjianto D, Djapic P, Teng F, Sturt A, et al. Strategic assessment of the
role and value of energy storage systems in the UK low carbon energy future. Rep Carbon Trust
Energy Futur Lab, Imp Coll EDF UK R&D Cent 2012:9.
[5] Drysdale B, Wu J, Jenkins N. Flexible demand in the GB domestic electricity sector in 2030.
Appl Energy 2015;139:281–90. doi:10.1016/j.apenergy.2014.11.013.
[6] Cheng M, Wu J. Availability of load to provide frequency response in the great Britain power
system. Power Syst … 2014.
[7] Office for National Statistics, “Households by household size and region of the UK”, [Online].
Available:
http://www.ons.gov.uk/ons/search/index.html?pageSize=50&sortBy=none&sortDirection=none&
newquery=households+in+London - Google Scholar n.d.
https://scholar.google.co.uk/scholar?q=Office+for+National+Statistics%2C+%E2%80%9CHouse
holds+by+household+size+and+region+of+the+UK%E2%80%9D%2C+%5BOnline%5D.+Avail
able%3A+http%3A%2F%2Fwww.ons.gov.uk%2Fons%2Fsearch%2Findex.html%3FpageSize%3
D50%26sortBy%3Dnone%26sortDirection%3Dnone%26newquery%3Dhouseholds%2Bin%2BL
ondon&btnG=&hl=en&as_sdt=0%2C5 (accessed December 21, 2015).
[8] Cheng M, Wu J, Galsworthy S. Power System Frequency Response From the Control of Bitumen
Tanks 2015.
[9] Short J, Infield D, Freris L. Stabilization of grid frequency through dynamic demand control.
Power Syst IEEE … 2007.

Biography
Meng Cheng received the B.Sc. degrees in electrical and electronic engineering from Cardiff
University, U.K. and North China Electric Power University (Beijing), China, in 2011, and the Ph.D.
degree from Cardiff University in 2015. She is currently a Research Associate with Cardiff University.
Her research interests include smart grids and dynamic demand.