2 Technical and Economic Aspects

2.1 Introduction

A good engineering project is generally a trade-off between a robust and effective design
and the cost of the final product. This involves the solution of a complex optimisation
problem, where the objective function is twofold: minimise the cost and maximise the
efficiency; the constraints are the equations that describe the physical behaviour of the
system; and inequalities, often binding, are technical and security limits.
CI-ESSs are, of course, no exception to this general procedure. In 2013, Stephen
Berberich of the California ISO, referring to the difficulties in building out grid-scale
energy storage, said: ‘It’s good stuff, but it’s expensive, and we have to find business
cases.’1 In the last 5 years the situation has substantially evolved and the number of
grid-scale projects has grown to a level at which it is safe to say that CI-ESSs are not just
an emerging trend but a solid market for the next decade(s). The illustrative examples
presented at the end of Chapter 3 strongly support this statement.
ESSs have an additional feature with respect to standard engineering projects and
products. There is, in fact, a sort of social acceptance of such systems, which goes be-
yond the mere economic benefit. This is so probably because energy storage is typically
coupled to eco-friendly renewable sources; or because trendy companies such as Tesla
are betting on this technology at many levels, such as electric vehicles (EVs), smart
houses and, lately, power electric systems. It is also true that any technology is expen-
sive when it is in its early stage and CI-ESSs are, de facto, still in their infancy. It is
unfair to judge energy storage technologies based on their current production costs. Our
attitude, as will be clear in the remainder of this chapter, is thus to assign a very small
weight, if any at all, to the part of the aforementioned ‘objective function’ that weights
cost minimisation.
This chapter focuses on technical and economic aspects related to ESSs. Section
2.2 discusses the technical applications of ESSs when included in transmission and/or
distribution networks and indicates the time scales of each application. The emphasis
is on the applications for dynamic analysis considered in Part III. Section 2.3 outlines
economic aspects of conventional and new energy storage technologies. This is the only
section of the book dedicated to economics. A motivation for this choice is also given
in this section.

1 Source: Green Tech Media, available at [Link].

32 Technical and Economic Aspects

2.2 Technical Aspects

Technical features, much more than cost and other economic considerations, allow a
proper classification of storage devices. With adequate incentives or long term invest-
ments, it is always possible to make a new technology worthwhile, but technical con-
straints are based on physics and, hence, not subject to the mood of the market. This
section explores in particular the main properties and specifications that quantitatively
characterise any storage technology (Section 2.2.1) and the services that the power grid
expects from energy storage (Section 2.2.2). In Section 2.2.2, the services for which
CI-ESSs are particularly suited are also highlighted.

2.2.1 Technical Specifications

The definitions and information included in this section were obtained from multiple
sources but mostly based on [119].

[Link] Capacity
Capacity is defined as the amount of energy available in the storage device when fully
charged. The energy retrieved when discharging the device is lower than the energy
stored in the device due to incomplete discharge and inevitable losses. It is thus con-
venient to distinguish between capacity, intended as the total energy required to fully
charge the storage device, Ech , and operational energy that can be utilised in practice,
Eop . The following relationship holds:
Ech ≥ Eop . (2.1)
These two quantities are energies and thus, in SI units, should be expressed in Joules.
However, it is more common to measure the capacity in Wh or multiples of it, e.g. kWh
and MWh. For batteries, it is also common to express the capacity in Amperes per hour
(Ah), i.e. in terms of the amount of storable electric charge. Eop is constrained by the
depth of discharge (see Section [Link]).
The ratio of Eop to Ech is not constant as Eop depends on the operation of the storage
device. A fast dynamic process (e.g. quick charge/discharge), can lower the efficiency of
some technologies (e.g. because of heating), thus leading to considerably decreased Eop .
In steady state, Eop decreases if the self-discharge is not null. Note also that, for some
technologies, Ech can change over time due to ageing or other factors. For example, the
capacity of a reservoir can decrease due to the accumulation of sediments.

[Link] State of Energy

The State of Energy (SoE) is defined as the available capacity and expressed as per unit
of its rated capacity:
E
SoE = , (2.2)
Eop
 2.2 Technical Aspects 33

where E is the amount of available energy stored in the device at a given time. For
batteries and electrochemical storage, it is more common to use the notation State of
Charge (SoC), which is effectively the potential energy of a battery.
The complement of the SoE (or SoC) is the Depth of Discharge (DoD), namely:
Eop − E
DoD = 1 − SoE = . (2.3)
Eop
For some technologies, the measure of the SoE or SoC is straightforward, e.g. for
flywheels, one just needs to know the inertia constant and the measure of the angular
speeds (see also Section 3.2.2). For other technologies, the determination of the SoE can
be involved and, as a consequence, not accurate. For batteries, for example, commonly
utilised methods are the voltage method (based on the discharge curve); the chemical
method (only for liquid electrolytes); and the
R current integration method, i.e. the indirect
measure of the electric charge as qe (t) = t i(r) dr. Kalman filtering is also utilised but
requires a precise model and, hence, precise parameters of the battery.

[Link] Power Rate

The power rate pn , often but not fully accurately referred to as maximum power, defines
the size of the energy storage device. Since a device does not always operate at the same
power, it can be expressed as an average quantity and is expressed in Watts or multiples
of it. The peak power is not necessarily the same as the maximum power, as it can be
associated with transient conditions, e.g. peak current, that can be sustained only during
a limited time, possibly shorter than the response time. Some ESSs can be characterised
by two pn , one for charge and one for discharge. For example, in compressed air energy
storage, the electrical machines connected to the turbine and the compressor can have
(slightly) different rated powers.
The power rate, along with the SoE, are often the only two quantities of storage
devices considered in the unit commitment problem. This is acceptable only if the dis-
charge time is way longer that the time period of each unit commitment. This is certainly
the case for conventional pumped hydroelectric and large compressed air energy stor-
age systems. The discharge time of these technologies is, in fact, of the order of weeks,
while the time period of the unit commitment is generally an hour or a fraction of it.
The simplistic formulation often utilised in the unit commitment problem should be
modified to take into account shorter discharge times of emerging technologies, which
are mostly based on power electronics, e.g. batteries.

[Link] Response Time

The response time T resp is the time required by the ESS to charge or discharge at rated
power from standby. T resp includes: (i) the time required by the ESS control system to
process the grid signal regulated by the ESS and received after a disturbance, T cont , and
(ii) the time for the ESS to ramp up or down from standby conditions to 100 ± 2% of the
ESS rated power, T ramp , as depicted in Figure 2.1 [37]. The time required to measure and
send the regulated grid signal to the ESS control system, T meas , must not be included in
the response time.
34 Technical and Economic Aspects

±2%
100

pn [%(MW)] T resp
T meas
T cont
T ramp

0
to Time [s]

Figure 2.1 Response time of an MW-sized ESS after a disturbance occurring at t = to .

[Link] Discharge Time

The discharge time is the time required to discharge, at the rated power, a fully charged
storage device:
Eop
T dis = . (2.4)
pn
The discharge time is a key design parameter of storage devices as their application and
system adequacy depend on it. It is often difficult to fully decouple the capacity and the
rated power, thus the optimum discharge time cannot always be easily identified.
For example, for superconducting magnetic energy storage, the capacity depends on
the current and the inductance of the coil, but the discharge time depends also on the
resistance and, hence, on the time constant of the overall circuit once the coil is dis-
charging (see also Section 1.2.3 on the discharge of a conventional iron-core coil). For a
pumped hydroelectric power plant, the capacity is proportional to the mass of water and
the height of the reservoir with respect to the turbine, but the discharge time depends
also on the size of the penstock and of the turbine.
For batteries, it is often preferred to express the discharge current or power for a
discharge time of 1 hour. This normalisation is useful as the capacity of the batteries
varies significantly from type to type and a direct comparison would be unfair without
proper normalisation. Two quantities are utilised, C-rate and E-rate. The C-rate indicates
the discharge current, in amperes, required to entirely discharge the battery in 1 hour.
For example, the discharge current of a 500 Ah 1 C-rate battery is 500 A. Similarly, the
E-rate refers to the discharge power of batteries. The E-rate is defined as the power to
entirely discharge the battery in 1 hour. For example, the discharge power of a 20 kWh
0.5 E-rate battery is 10 kW.
Figure 2.2 illustrates the rated power versus the discharge time of a variety of en-
ergy storage technologies. The discharge times of existing storage technologies span
 2.2 Technical Aspects 35

Demand Transmission & distribution Supply

Power to hydrogen and to SNG

Months
Fuel cells

PHES
Days

Flow battery
C-CAES
Discharge time
Hours

NaS
MSTES
Ultrabattery CAES
LAES
Lithium ion
Minutes

FES
Lead acid
NiCd

NiMH
Seconds

SMES
ECES

10 kW 100 kW 1 MW 10 MW 100 MW 1 GW
Power rate

Figure 2.2 Power rate vs discharge time of a variety of energy storage technologies. The meaning
of the acronyms is given in the notation section in the preamble of the book as well as in Chapter 3.

several orders of magnitude, from tens of milliseconds, e.g. supercapacitors, to months,

e.g. power-to-hydrogen technology. The vast majority of the technologies, while pos-
sibly spanning a few orders of magnitude in terms of power rate, cover only a specific
time scale. This explains why, in practical applications, different technologies are cou-
pled together in hybrid solutions, e.g. a bank of batteries with long discharge time and
high power rates, but also slow response times, is coupled with other technologies such
as flywheels or supercapacitors, which show short discharge times but also fast response
times.

[Link] Efficiency
The round-trip efficiency or, simply, efficiency, is defined as the ratio of the operational
usable energy to the stored energy when the storage device is fully charged:
Eop
η= . (2.5)
Emax
The efficiency is a key parameter of any storage technology as it heavily impacts on
its economic viability. The calculation of the efficiency is often not straightforward
36 Technical and Economic Aspects

because Eop is a function, as discussed above, of the operating conditions, duty cy-
cle and dynamic response of the storage device. In the long term, self-discharge and
ageing also modify the efficiency. All these factors suggest that the efficiency is a time-
dependent quantity that should be defined per cycle and that there is a discharge time
and a charge/discharge power that maximise the efficiency [119]. The calculation of
such an optimum is a relevant aspect that should be taken into account in the economic
appraisal of storage devices for grid applications.
Let us consider a simple example. The energy needed to pump a water volume V to a
height h is:
ρghV
Epump = , (2.6)
ηpump
where ρ is the density of the water, g is the gravitation acceleration and ηpump is the
efficiency of the pump. On the other hand, the energy obtained from the water stored in
the reservoir through the turbine is:
Eturb = ρ g h V ηturb , (2.7)
where ηturb is the efficiency of the energy generation. Assuming no leakages, the round-
trip efficiency of the energy storage system is thus:
Eturb
η= = ηturb ηpump . (2.8)
Epump
For pumped hydroelectric energy storage, η falls in the range from 70 to 85%. Indeed, a
similar expression can be obtained for all storage technologies, i.e. the overall efficiency
is given by the product of the efficiencies of the charging and discharging processes,
namely:
η = ηch ηdis . (2.9)
In some cases, e.g. batteries and supercapacitors, energy leakage (i.e. self-discharge)
is not negligible and Equation (2.9) must take into account the time during which the
energy is stored and the self-discharge energy. Table 2.4 later on in this chapter provides
data on the efficiency of most relevant storage technologies.

[Link] Specific Energy and Energy Density

The specific energy is the maximum amount of stored energy, Ech , per unit of mass,
m, of the storage unit, i.e. Ech /m, which is expressed in Joule/kg or, more commonly,
W h/kg. Alternatively, instead of the specific energy, one can define the energy density
as the stored energy per unit of volume. The energy density is handy if the density of
the matter utilised for the storage is variable, e.g. compressed air technology.
The specific energy is particularly relevant for portable applications, e.g. EVs and
aerospace. While less crucial for permanent installations, the specific energy is still
often relevant for economic considerations.
 2.2 Technical Aspects 37

[Link] Specific Power and Power Density

These quantities are defined as the specific energy and energy density, respectively, that
can be delivered or absorbed by the storage device in a unit of time. These are relevant
quantities in EV applications (e.g. acceleration).

[Link] Autonomy
Autonomy is defined as the maximum period of time during which a storage device can
continuously release energy, as follows:
Eop
Ta = , (2.10)
pdis
where pdis is the average power during the discharge. The autonomy is conceptually
very similar to the discharge time, and differs from it because pdis , pn , in general.
Autonomy is particular relevant for applications where the storage device has to operate
disconnected from the grid, such as EVs, and when coupled with intermittent energy re-
sources. Particularly relevant for grid applications is the combination of thermal storage
with solar power plants (see the real-world examples described in Section 3.7).

[Link] Control Capability

The control capability depends on the capacity, power rate and response time. This spec-
ification covers several applications and thus requires a thorough discussion that goes
well beyond a simple definition. Details on control capabilities of ESSs are provided in
Section 2.2.2.

[Link] Operational Cost

The operational cost depends on many technical factors, including cost of materials and
maturity of the technology, but also on political and social factors, e.g. state incentives to
support a given technology, as happened in the past for wind and PV solar installations.
A simple expression to calculate the operational cost of an ESS is as follows:
C = c1 Eop + c2 pdis , (2.11)
where c1 is the unit cost per operational energy capacity; and c2 is the unit cost per
discharge power [119]. Equation (2.11) can be rewritten as:
C = (T a c1 + c2 )pdis , (2.12)
where T a is the autonomy defined in Equation (2.10).
Investment costs are more complex to calculate as they require a projection through
the lifespan of the system, e.g. 20 years for typical electricity production systems. To be
able to compare different technologies, it is also desirable to define a normalised cost.
All these aspects are thoroughly discussed in Section 2.3.

[Link] Cycling Capacity

The cycling capacity, also called durability, is the maximum number of round-trip cycles
for which the storage unit can release its rated energy. In some references, the durability
38 Technical and Economic Aspects

is quantified as the State of Health (SoH) of the ESS, especially referred to batteries.
SoH does not generally refer to a specific physical quantity. It is determined heuristically
(and often treated as a trade secret) based on a combination of several parameters such
as internal resistance, self-discharge, and the number of completed cycles.

[Link] Self-discharge
Self-discharge is the quota of energy that is dissipated after charging an ESS during the
time the system is not utilised. Depending on the technology, the self-discharge can be
negligible or even null, e.g. an energy system based on potential energy (see for example
the solid-masses ESS described in Section 3.7), or substantial, such as electrochemical
capacitors with non-negligible internal resistance.

[Link] Thermal Runaway

Thermal runaway is an issue of certain battery technologies and consists in self-acce-
lerating temperature rise of the cells. The high temperature of the electrolyte triggers
heat-generating exothermic reactions. Such reactions further increase the temperature
of the electrolyte. The process can lead to disruption and even to fire or explosion,
unless the battery is properly cooled down. For example, the thermal runaway process
can initiate in lithium-ion batteries in a range from 365 to 395 K [17], and from 325 to
335 K for valve regulated lead-acid batteries [80].

[Link] Ageing
Ageing is the deterioration of the performance and efficiency due to time. This affects
all systems (both mechanical and organic) and energy storage devices are no excep-
tion. Mechanical energy storage ages mainly due to fatigue and wear, chemical storage
because of oxidation and electromagnetic storage because of the deterioration of the
insulation, e.g. thermal degradation of the polymers or partial discharges.
In any case, ageing is related to the cycling capacity, the average SoE and the energy
per cycle. Since it is based on statistical evaluations, the lifetime of a device can be
defined only approximately.

2.2.2 Grid Applications

The main applications of ESSs are related to ancillary services and regulation for trans-
mission and distribution systems. These span several time scales, as expected from the
discussion of Figure 2.2. The following applications are discussed, from the slowest
to the fastest: seasonal storage, arbitrage, T&D congestion relief and infrastructure in-
vestment deferral, off-grid utilisation, black start, non-spinning reserve, load levelling,
load following, spinning reserve, integration of intermittent resources, frequency con-
trol, voltage support and RoCoF control.
Some of these applications, such as seasonal storage and load levelling, have been in
place for decades and are mostly based on conventional pumped hydroelectric power
 2.2 Technical Aspects 39

plants. Others, in particular RoCoF control, off-grid utilisation and integration of inter-
mittent resources, are recent developments and mostly based on CI-ESS. The latter is
also a crucial aspect of the microgrid and smart house concepts (see Section 4.5.2).
Typical energy storage applications are shown in Table 2.1, which is mainly based on
data collected by the International Energy Agency [130]. Figure 2.3 shows the typical
time scales and capacities of the grid applications for different energy storage technolo-
gies.

Table 2.1 Relevant parameters of power system applications of energy storage systems
(Source: International Energy Agency).

Capacity Discharge
Application Cycles T resp
(MW) duration

Seasonal storage 500–2,000 days–months 1–5 p/year >day

Arbitrage 100–2,000 8–24 hours 0.25–1 p/day >1 hour
T&D congestion 10–500 2–4 hours 0.14–1.25 p/day >1 hour
relief
T&D infrastructure 1–500 2–5 hours 0.75–1.25 p/day >1 hour
investment deferral
Off-grid 0.001–0.01 3–5 hours 0.75–1.5 p/day <1 hour
Black start 0.1–400 1–4 hours <1 p/year <1 hour
Non-spinning reserve 10–2,000 15 min–2 hours 0.5–2 p/day >15 min
Load levelling 1–2,000 15 min–1 day 1–29 p/day >15 min
Spinning reserve 10–2,000 15 min–2 hours 0.5–2 p/day <15 min
Integration of 1–400 1 min–hours 0.5–2 p/day <15 min
intermittent resources
Frequency regulation 1–2,000 1–15 min 20–40 p/day 1 min
Voltage support 1–40 1 s–1 min 10–100 p/day 0.1–10 s
RoCoF control 1–2,000 1s 10–100 p/day 0.1–5 s

[Link] Seasonal Storage

Seasonal storage refers to the ability to store energy for long periods, from days to
months. This has been the main historical purpose of the reservoirs of hydroelectric
power plants. In recent years, fuel cells and power to hydrogen and to synthetic natural
gas are seen as excellent candidates for long-term seasonal storage, whereas flow batter-
ies and compressed air energy storage appear adequate for the time scale of days. There
exists also a thermal seasonal storage which consists in storing heat in the summer and
re-utilising it in the winter. This kind of thermal storage utilises underground energy
storage systems, which, in some way, are the evolved version of traditional cellars.
40 Technical and Economic Aspects

Demand Transmission & distribution Supply

Months
Seasonal
storage
Days

Waste Load Arbitrage

Demand shifting
and Peak Reduction levelling

Combined heat power

Discharge time
Hours

Offgrid
Black start
T&D relief & deferral Spinning
and
non-spinning
Variable supply reserve
resource
Minutes

integration
Primary frequency control Secondary
frequency control
Seconds

RoCoF control Voltage support

10 kW 100 kW 1 MW 10 MW 100 MW 1 GW
Capacity

Figure 2.3 Power requirement versus discharge duration for power system applications.

[Link] Arbitrage
Arbitrage refers to the energy trade among energy markets and consists in acquiring
and storing energy in periods of low price (overnight and weekends) and then selling
the energy in periods when the energy has a high price (daytime of weekdays). An ex-
ample of systematic arbitrage is the utilisation of pumped hydroelectric storage in Italy,
which stores the cheap energy produced in France by nuclear power plants overnight
and resells the energy to France at peak hours. The time scales involved in arbitrage
are days and weeks, and the technologies are the same as those discussed for seasonal
storage above.

[Link] T&D Congestion Relief and Infrastructure Investment Deferral

During the operation of transmission and distribution systems, it is often required to
dispatch relatively more expensive generation to avoid line congestion. This can be
avoided if ESSs with a discharge time of a few hours or more are available in the system.
For the same reason, ESSs reduce the need for building new transmission lines where
these are required only to alleviate congestion during a few hours of weekdays. For
the time constants involved, congestion relief can be obtained not only with pumped
 2.2 Technical Aspects 41

hydroelectric and conventional compressed air ESSs, but also with large battery banks
and converter-interfaced compressed and liquid air ESSs.

[Link] Off-grid Operation

The ability to operate off-grid, i.e. fully disconnected from the grid is one of the prop-
erties that make CI-ESSs – batteries, in particular – enablers of new solutions, such as
microgrids, smart houses and EVs (see also Section 4.5.2). The autonomy of the off-
grid operation is proportional to the capacity of the storage device, which, ultimately, is
constrained by the cost of the storage technology. Currently, the time scale of off-grid
operation is about one hour.

[Link] Black Start

Black start is the ability of a power plant to start up without the need for external energy
sources. This ability is crucial after a major blackout to re-energise the power system.
Energy storage systems that are able to operate off-grid are thus ideal candidates for
this important, although very rarely needed, service. Unless the blackout is a conse-
quence of major damage to the transmission system, e.g. as a consequence of extreme
weather conditions, the time scale of black start is about an hour, i.e. the time required
by conventional generation to start-up and synchronise with the system.

[Link] Load Levelling

Load levelling, or peak shaving, consists in reducing the peaks and increasing the val-
leys of the daily load profile. It is a process that spans 24 hours and is similar, from
the storage and market points of view, to arbitrage. The reader can refer to the example
discussed in Section 1.2.7 that illustrates the typical load levelling effect of large energy
storage systems.
Load levelling should not be confused with load following, which consists in balanc-
ing the demand during the normal operation of the power system. Along with base load
– i.e. constant maximum power – and peaking, load following is an operation mode of
conventional power plants. While load levelling makes sense for energy storage due to
its cyclical behaviour, load following is specific to the scheduling of conventional gen-
eration. However, if the time scale is short enough, load following effectively overlaps
with the SFC, which can be provided by energy storage devices (see the case study
discussed in Section [Link]).
The storage technologies that can be utilised for load levelling and SFC are the same
as those mentioned for network congestion relief.

[Link] Integration of Intermittent Resources

The power smoothing of non-dispatchable energy resources such as wind and solar is
currently one of the most popular applications of energy storage devices (some real-
world examples are described in Section 3.7). Whether it is better to centralise or dis-
tribute storage devices for this kind of application is still an open question (see also the
discussion in Section 2.3 below). In any case, the time constants associated with power
smoothing vary from a few seconds to several tens of minutes, even up to some hours,
42 Technical and Economic Aspects

and it thus typically requires hybrid solutions that mix together ESSs with fast response
time (e.g. electrochemical capacitor, superconducting magnetic and flywheel ESSs) and
others with longer response time (e.g. battery ESS).

[Link] Spinning and Non-spinning Reserve

The ability to compensate the sudden unscheduled disconnection of a generator is called
‘reserve capacity’ or ‘frequency containment reserve’. Depending on the time scale,
the reserve is classified as spinning for response times shorter than 15 minutes; and
non-spinning for response times longer than 15 minutes. From the energy storage de-
vice point of view, the main aspect is not the time response, which can be even faster
than typical spinning reserve, but rather the availability of a sufficient capacity to cover
the imbalance created by the generator outage. Energy storage devices can serve as a
‘buffer’ to allow starting up slower technologies (e.g. gas turbines).

[Link] Frequency Control

The PFC and SFC refers to the continuous balance, in the time frame from a few seconds
to a few minutes, of load variations. Examples of conventional PFC and SFC are given in
Sections 1.2.5 and 1.2.6, respectively. Moreover, the case studies discussed in Chapters
7 and 8 illustrate the ability of CI-ESSs to provide PFC and, to a lesser extent, SFC, as
well as the impact of such controls on power system dynamics.

[Link] Voltage Support

Most ESSs can provide voltage control. Conventional pumped hydroelectric and com-
pressed air ESSs include a voltage controller that regulates the field voltage of the syn-
chronous generators. CI-ESSs have a voltage support capability thanks to the VSC reg-
ulation of the voltage on the AC side of the converter. In fact, a VSC alone without
energy storage can be utilised as a Flexible AC Transmission System (FACTS) device
(e.g. Static Synchronous Compensator (STATCOM)). The time scale of voltage support
is of the order from hundreds of milliseconds to a few tens of seconds. Voltage support
and stability through CI-ESSs is further discussed in Chapter 7 and in Section 9.3.

[Link] RoCoF Control

As discussed in the example given in Section 1.2.4, synchronous machines, through
their inertia, naturally take care of balancing energy in the very first instants after a
variation of demand or supply. With the massive introduction of CIG, however, the
level of inertia of power systems has considerably decreased and can vary substantially
during the operation of the system. It is thus required that non-synchronous generation
as well as storage provide the response, namely RoCoF control, that in the past was
provided by synchronous generators. This topic is further discussed in Section 4.4. The
interested reader can also find in [190] an in-depth survey of the dynamics and control
of low-inertia systems.
 2.3 Economic Aspects 43

2.3 Economic Aspects

ESSs are generally costly. Except for pumping hydroelectric power plants, which have
been exploited since the 1930s (see Figure 3.2) and very peculiar and rather unique stor-
age solutions (see Section [Link]), energy storage has never really been economically
viable. It is also undeniable that energy storage is the solution to most power system
problems as thoroughly discussed in the motivations and technical challenges earlier in
this chapter. As a matter of fact, if a consistent amount of fast-controlled large-capacity
ESSs were installed in a power system, most stability and operational problems would
be finally solved.
It is a matter of research and investment to find economically viable energy storage
systems.2 Actually, in hindsight, it is strange that the research on cheap storage sys-
tems has not begun earlier. The reason is, likely, the profound conservatism of system
operators, which has had to drastically change in recent years due to deregulation first,
and then to the strong social and political pressure for the reduction of pollution. As
discussed in Section 2.2, without the trigger of the high penetration of renewable re-
sources, and if such resources were not mostly stochastic, the time for energy storage
would have very likely waited a long while.
On the other hand, in many, even non-technical news media, energy storage – batter-
ies, in particular – is seen as the holy grail of power systems and the key for moving
towards the long sought goal of fully clean energy.3 The cost to store a kWh of electric
energy is key. If this cost is low enough to make it more convenient to install storage
devices than additional generators to cope with the instantaneous power balance, then
the operation of the power system will change forever (for the technicalities, the reader
can refer to Section 2.2). It is also worth noticing at this point that the drastic reduction
in battery costs and increase in performance are expected to continue in the near future,
at least for batteries based on lithium ion, sulphur and vanadium [133]. This will be
achieved by increasing the energy density of the batteries.
So, given that the cost of a kWh is the deal-breaker at this time for a massive pen-
etration of converter-interfaced energy storage devices, the remainder of this section
provides some figures to illustrate the trend of the most important technologies and the
projections (inevitably tentative) for the next decade.
At this time, a 100 $/kWh is assumed to be the threshold below which batteries will be
more competitive than internal combustion engines and, thus, the entire industry of EVs
would be revolutionised. In 2013, Tony Seba, entrepreneur and solar power expert from
Stanford University, stated: The tipping point for the mass market to move from internal
combustion engines to EVs is between 250 and 300 $/kWh. Once it gets to 100 $/kWh,
it is all over. I think we will get to 250 $/kWh by 2020. By 2030, when batteries are

2 In 2016, the US Energy Department was funding 75 projects focusing on the development of energy
storage devices, including research groups from MIT, Harvard, Stanford, and Lawrence Livermore and
Oak Ridge National Laboratories. (Source: The Telegraph, 10 August 2016).
3 The New York Times was among the first to utilise the expression ‘holy grail’ referred to energy storage, in
an article of 6 November 2010, available at: [Link]/2010/11/07/us/[Link].
44 Technical and Economic Aspects

at 100 $/kWh, gasoline vehicles will be obsolete. Not on their way out, obsolete.4 This
statement also assumes that the price per kWh of combustion engines is not expected to
decrease in the future.
The threshold value for power electric systems is not a straightforward quantity to
be computed, as technologies are very diverse, and cost is not always the primary pa-
rameter to be taken into account. For example, one of the cheapest, most reliable and,
in many aspects, cleanest way to produce electricity is through nuclear fission but, in
most developed countries, especially Western culture ones, this technology is not an
option anymore.5 In any case, while a threshold value is not attempted to be defined in
this book, it is a fact that the installation of grid-connected energy storage devices other
than PHES ones has more than quadrupled in the last decade (see Figure 3.3), which is a
clear sign that there is a strong interest in energy storage for power system applications.
A widely accepted quantity to estimate and compare diverse technologies related to
energy generation is the Levelised Cost of Electricity (LCoE), or Levelised Energy Cost
(LEC), which is a measure of lifetime costs divided by energy production. The LCoE is
utilised as an approximated estimation of the average price that a given technology has
to obtain in a market to break even over its lifetime. Lifetime costs typically include the
total investment, operations and maintenance, fuel, decommissioning and treatment of
waste.
Since the LCoE is a cost normalised with respect to the unit of energy produced
by the power plant, it allows comparing diverse technologies (e.g. wind, natural gas,
battery) of diverse life spans, size, capital cost, risk, return and capacities. The LCoE is
calculated over the design lifetime of a plant, which is usually 20 to 40 years, and given
in the units of currency per kWh or multiples of it.6
A simple expression for the LCoE is the following [131]:
PT
(it + mt + ft + wt ) · Et · (1 + %)−t
LCoE = t=1 PT , (2.13)
t=1 E t · (1 + %)
−t

where:

T: expected lifetime of the system in years;

%: discount rate;
it : total investment and capital construction cost in year t in $/MWh;
mt : cost of operation and maintenance in year t in $/MWh;
ft : fuel cost in year t in $/MWh;
wt : cost of decommissioning and waste in year t in $/MWh;
Et : electrical energy generated in year t in MWh.

The quantity (1 + %)t is the discount factor or time-value for money in year t.
Several variants of (2.13) have been proposed. In [45], the formula also includes the
4 See full article at: [Link]
5 A noteworthy, well-known exception is UK, where EDF is currently building a new nuclear power station
at Hinkley Point C in Somerset.
6 An interactive web-based LCoE calculator is available at: [Link]/analysis/[Link]
 2.3 Economic Aspects 45

depreciation schedule rt and the salvage value st of any physical assets at the end of
life-cycle:
PT
(it + mt + ft + wt − rt ) · Et · (1 + %)−t − st · Et · (1 + %)−T
LCoE = t=1 PT . (2.14)
t=1 E t · (1 + %)
−t

The calculation of the LCoE is illustrated with a simple numerical example. Consider
a nuclear power plant with net capacity Pnet = 1, 400 MW, and assume the following
values based on [131]:

%= 3% ;
T= 60 years ;
it = 30.75 $/MWh ;
mt = 11.0 $/MWh ;
ft = 11.33 $/MWh ;
wt = 1.26 $/MWh ;
rt = St = 0 .

The electrical energy generated in a year is the average hourly power production, ph ,
multiplied by the number of hours in a year:
Et = ph · 8,760 . (2.15)
Since nuclear power plants are operated in such a way that the production is as con-
stant and as close as possible to the rated power of the plant, it can be assumed that
ph ≈ pnet , hence:
Et ≈ pnet · 8,760 ≈ const., ∀t = 1, . . . , n . (2.16)
Then, since all costs are constant, (2.13) simply becomes:
LCoE = it + mt + ft + wt , (2.17)
and, substituting the values of the example:
LCoE = 30.75 + 11.0 + 11.33 + 1.26 = 54.34 $/MWh .
Calculations given in [131] are based on (2.17) rather than on (2.13) or (2.14). The
costs and the power production, in fact, are assumed constant for all technologies and
are thus independent of the net capacity of the technology and of the discount rate %.
In [131], the LCoE varies as % varies because it and mt are different for different %.
The calculation of lifetime costs might be biased by several external and often non-
technical factors, e.g. taxes or subsidies, and other, more sophisticated versions of (2.13)
have been proposed. Some of them are specific for ESSs, e.g. [224], which includes the
relations between storage and PV generation. The financial advisory and asset man-
agement firm Lazard conducts an annual evaluation of the Levelised Cost of Storage
(LCoS) which includes most operational parameters of battery storage systems.7 More-
over, the numerical example above shows that, even assuming that the LCoE formula
7 The studies are available at: [Link].
46 Technical and Economic Aspects

is accurate, it is quite difficult to have a precise estimation of the costs through the life-
time of the power plant. Very often the LCoE is thus calculated only based on present
information and extrapolating it to the next decades.
The standard definition of LCoE does not take into account operational aspects either.
Among these, we cite dynamic response, i.e. the ability to follow fast varying demand or
to match the demand at all. This is a major aspect of all stochastic renewable generation,
which may lead to a need for extra generation backup. In [289], the authors conclude
that replacing nuclear power with wind-based generation would lead, at least in Sweden,
to double costs and emissions. The intermittency and volatility of wind and solar power
have a significant associated ‘cost’, which should not be neglected when evaluating the
actual LCoE.
Other authors believe that the intermittency can be smoothed (statistically) if a suf-
ficiently large and geographically extended network is considered (see for example,
Chapter 2 of [2]). This argument, however, is not convincing. In a small system, where
distances are short, it is unlikely that weather conditions are so uniformly distributed
that the average wind and/or solar power production is constant. In Ireland, for exam-
ple, the volatility of wind power generation is quite high even in a short time, and it is
not uncommon to have days during which the wind power production is null. On the
other hand, for large systems, the statistical argument given in [2] does not take into
account the location of the generation. For example, considering the European inter-
connected system, ENTSO-E, it is not the same to have all power generated by solar
power plants, say, in Spain, while the rest of continental Europe is under the clouds, as
compared with the same amount of generation being evenly distributed everywhere in
the system. Statistical considerations have still to satisfy the Kirchhoff laws of the elec-
tric network and stability constraints. In any case, whatever one’s belief, cost effective
and fast responding ESSs are to be sought.
There is another and deeper reason that justifies why the LCoE should be utilised with
caution. As any quantity that mixes a cost or a price with an actual physical quantity –
in this case, energy – the LCoE embeds an intrinsic arbitrariness. The costs that appear
in (2.13) are, in fact, decided based on criteria that do not follow physical principles or
technical constraints but, rather, political and social situations. As such, the LCoE does
not have an absolute significance. Only indexes that are exclusively based on physical
quantities, e.g. the efficiency, can provide valuable and consistent information. If the
price of an object, in fact, were defined as the amount of energy required to make it
and were measured in joules, there would be no market speculation or uncertainty. It is
the ‘human’ – artistic, sentimental, speculative, etc. – value attributed to an object that
makes prices and, hence, market dynamics largely unpredictable.
The LCoE ability to indicate the relative significance of different technologies is
doubtful, as political and social beliefs tend to change in time, and do not always follow
a rational economic criterion.8 The huge international effort to reduce CO2 emissions
and pollution is a good example of that and, ultimately, the reason why energy storage
devices have suddenly become so relevant in the last decade, despite their high costs.

8 The tulip mania that occurred in the Netherlands in the seventeenth century should always be kept in mind.
 2.3 Economic Aspects 47

With the caveats above in mind, Table 2.2 shows the expected LCoEs of most relevant
generation types for power plants commissioned in 2020, as published by the Interna-
tional Energy Agency (IEA) and Nuclear Energy Agency (NEA) in 2015 [131]. The
LCoEs included in the table consider 19 OECD (namely, Organisation for Economic
Co-operation and Development) and three non-OECD countries.
The estimation of cost depends on many factors which can be weighted differently
by different agencies. Table 2.3, shows the estimated LCoEs for new generation re-
sources commissioned in 2022 as published by the US Energy Information Administra-
tion (EIA) in 2017 [279]. The values in the table only refer to the United States. Note,
for example, the different figures for nuclear power in Tables 2.2 and 2.3.
The figures for conventional and renewable generation types serve as a comparison
base for better understanding the cost of energy storage technologies. Table 2.4 shows
typical ranges of the power rate (pn ), efficiency (η), lifetime (T ) and LCoE of most
common energy storage technologies in 2017 as provided online by Enipedia.9

Table 2.2 Expected LCoE ranges in $/MWh of commissioning most relevant generation tech-
nologies in 2020 at 3% discount rate (Source: IEA and NEA).

Low High
Technology
($/MWh) ($/MWh)
Natural gas-fired CCGT 61 133
Coal 66 95
Nuclear 29 64
PV solar – residential 96 218
PV solar – commercial 69 142
PV solar – large, ground-mounted 54 181
Onshore wind 33 135
Offshore wind 98 214

Table 2.3 Expected LCoE ranges in $/MWh of commissioning most relevant generation tech-
nologies in 2022 (Source: EIA).

Weighted average Simple average

Technology
($/MWh) ($/MWh)
Natural gas-fired CCGT 58.6 57.3
Coal – 123.2
Nuclear 96.2 99.1
PV solar 73.7 85.0
Geothermal 44.0 46.5
Biomass 97.7 102.4
Thermal solar – 242.0
Onshore wind 58.8 63.7
Offshore wind – 157.4
Hydroelectric 63.9 66.2

9 The website is available at: [Link] Storage Technologies.

48 Technical and Economic Aspects

Table 2.4 Power rate, efficiency, lifetime and LCoE of most common energy storage tech-
nologies in 2017 (Source: Enipedia, TU Delft).

pn η T LCoE
Technology
(MW) (%) (years) ($/kWh)

Advanced compressed air 1–1,000 55–70 50 0.008–0.015

Compressed air 1–300 29–43 50 0.02–0.04
Edison (NiFe) battery 0.001–0.01 <65 50 0.016–0.13
Electrochemical capacitor 10 90 10–20 0.02–0.2
Flow batteries 0.01–10 70 5–10 0.05–0.8
Flywheels 0.001–10 84 15–30 0.257–0.92
Hydrogen storage 0.1–10 20–30 20 0.20–0.35
Lead-acid battery 0.001–10 65–85 5–15 0.20–1.2
Lithium-ion battery 0–10 86–96 12–15 0.4–1.0
Nickel-cadmium battery 30 80 20 0.20–1.0
Pumped hydroelectric 100–1,000 70–85 60 0.01–0.015
Salt-water (sodium-ion) battery 320 85 20 0.2–0.5
Sodium-sulphur battery 30–35 80–85 15 0.05–0.14
Superconducting magnetic coil 0.1–10 >90 20 72.5
Vehicle-to-grid demand response 0.001–0.03 63–72 5 0.5–1.2