You are on page 1of 189

POWER SUPPLY

CHALLENGES

Solutions for Integrating Renewables


"A timely book to demonstrate how to balance the power supply
and demand under very complex conditions, such as renewables
integration, temperature variation and demand uncertainty."

Chongqing Kang
Professor
Department of Electrical Engineering
Tsinghua University
China

"The author does a nice job to describe the changing character of


electric power generation while respecting the fundamentals and
working principles of an integrated grid."

Thomas Key
Senior Technical Executive
Electric Power Research Institute EPRI
USA
Solutions for Integrating Renewables
Solutions for Integrating Renewables Jacob Klimstra
Table of Contents:
Foreword............................................................................................................................................... 8
Note to the reader............................................................................................................................... 9

1. How to secure the electricity supply in a changing world........................................ 10


1.1. An affordable, reliable and sustainable supply of electricity...................................... 12
1.2. Challenges of renewable energy sources........................................................................ 18
1.3. Faults and failures in electricity supply systems........................................................... 21
1.4. Balancing supply and demand in energy markets........................................................ 22
1.5. Conclusions........................................................................................................................... 26

2. Balancing the electricity supply in case of calamities......................................................28


2.1. Matching electricity supply and demand....................................................................... 30
2.2. Primary control reserves compensating for the failure of a power plant................ 31
2.4. Conclusions........................................................................................................................... 49

3. Balancing power demand and supply when conditions change....................................50


3.1. Electricity supply differs depending on local situations............................................. 52
3.2. Power demand pattern in Finland exemplifying an industrialised nation.............. 52
3.3. Power demand in the Republic of Ireland, exemplifying a system with
much wind-based power.................................................................................................... 56
3.4. The 50Hertz transmission system operator region in Germany,
a region with much solar-based power........................................................................... 67
3.5. Effects of photovoltaics on other power plants in Texas and California................ 72
3.6. Conclusions........................................................................................................................... 75

4. Active and reactive power............................................................................................ 76


4.1. Reactive power analogies................................................................................................... 78
4.2. The three basic load elements in alternating current systems................................... 78
4.3. The power factor cos φ....................................................................................................... 84
4.4. Impedance of electricity transmission systems............................................................. 86
4.5. Voltage change over a power transmission line............................................................ 89
4.6. Risks created when insufficient reactive power is supplied
by renewable energy sources............................................................................................ 93
4.7. Conclusions........................................................................................................................... 95

© Jacob Klimstra & Wärtsilä Finland Oy Publisher: Wärtsilä Finland Oy


Editorial work: Jussi Laitinen 1st edition
Graphic design: Jiipee Mattila ISBN 978-952-93-3634-0
Printing house: Arkmedia, Vaasa 2014 ISBN 978-952-93-3635-7 (pdf)
5. Energy storage............................................................................................................... 96
5.1. The enormous challenge of energy storage................................................................... 98
5.2. Basic properties of energy storage devices...................................................................100
5.3. Applications for energy storage devices.......................................................................101
5.4. Methods and costs of energy storage............................................................................104
5.7. Discussion on energy storage..........................................................................................117
5.6. Conclusions.........................................................................................................................119

6. Costs of producing electricity...................................................................................120


6.1. Challenges in determining kWh costs..........................................................................122
6.2. Varying conditions for generating electricity..............................................................122
6.3. Cost analysis for different generating techniques.......................................................124
6.4. The total costs of producing electricity........................................................................134
6.5. The electricity price for consumers...............................................................................139
6.6. Discussion regarding electricity production costs.....................................................140
6.7. Conclusions.........................................................................................................................140

7. Future power supply systems.....................................................................................142


7.1. The road towards an optimum power supply system................................................144
7.2. An optimised generating portfolio without renewables...........................................145
7.3. An optimised power plant portfolio design with renewable energy sources.......149
7.4. Discussion regarding the suggested optimum power supply portfolio.................157
7.5. An optimum electricity generation portfolio for emerging economies................159
7.6. Conclusions.........................................................................................................................160

8. Power supply challenges – A review.........................................................................162


8.1. Realism is needed in the energy debate.......................................................................164
8.2. Low capacity factors escalate balancing issues............................................................165
8.3. Flexible local generators of limited size offer excellent backup for renewables......167
8.4 Natural gas is ideal for backup capacity.........................................................................167
8.5. Integrating power demand and heat demand offers good perspectives................168
8.6. Agile, flexible power plants help to ensure a reliable
and cost-effective power supply.....................................................................................169
8.7. Outlook for the future......................................................................................................170

Appendix 1.......................................................................................................................................172
Appendix 2.......................................................................................................................................176
Biograph............................................................................................................................................182
References........................................................................................................................................183
Glossary............................................................................................................................................185
Foreword
Hans ten Berge
Secretary General of Eurelectric

Europe's electricity markets are changing. Historically, electricity markets were based on genera-
tion capacity with comparatively low fixed costs and high variable (fossil) fuel costs. But the ratio
of variable and fixed costs is shifting, as renewable generation based on solar and wind with little
to no variable cost increasingly enters the market.
Despite technological advances and efficiency gains, even the most modern, state-of-the-art
fossil fuel capacity is finding it increasingly difficult to compete in a market where subsidised
capacity is able to generate at zero variable cost. Yet firm capacity will be needed to back up
variable generation. In this context, many argue the current market environment is no longer
fit for purpose.
The penetration of low-carbon, intermittent, generation capacity is a positive step towards
less carbon-intensive electricity systems. However, the change in generation portfolios does
pose a number of challenges for the markets. How do we ensure that sufficient capacity is
available when the wind doesn't blow and the sun doesn't shine? Currently subsidy schemes
remove variable renewable capacity from market price signals. This cannot and should not be a
long-term option – for any type of electricity generation.
This is no academic debate. Rather, the European electricity industry is already feeling
the effects of the recent changes on the ground. Companies have to mothball recently built
gas plants, for instance, or put investment projects on hold. The unfavourable market condi-
tions are also discouraging external investors from putting their money into the electricity
sector. Meanwhile, policy support costs, for instance for renewables or energy efficiency, are
increasing the price that end customers pay for their electricity.
In short: the challenges are big and the solutions are as yet unclear. A fresh look at market
design and at a 'smarter' energy system in general is needed. This book contributes to that dis-
cussion, with a particular focus on the effects for generators. In describing and analysing the
current environment and the way that some of the challenges can be addressed, it addresses
key concerns of the European – and global – electricity industry today.
Note to the reader
Jacob Klimstra

Many readers of this book have to make important decisions about electrical energy supply in
different regions. Electricity is crucial for creating wealth and comfort in a modern society. With
ever-growing demand for electrical energy, its generation has a huge impact on global fuel con-
sumption and the related emissions. This demand is expected to double over the next twenty years
or so.
During the past hundred years, scientists and engineers have acquired a thorough knowl-
edge of the power supply system. Modern power plants achieve high fuel efficiencies, and
their emissions have been drastically reduced. Excellent transmission and distribution systems
ensure a high degree of reliability in the power supply to consumers. However, measures need
to be taken to make the electricity supply more sustainable. Ultimately, the depletion of fossil
fuels will occur and the issue of global warming from greenhouse gases cannot be neglected.
Therefore, the power sector has to adjust accordingly and find a new path. Decisions made
now will have a long-term impact and optimum solutions have to be chosen.
The purpose of this book is to explain the challenges arising from the advent of a large
volume of intermittent renewable energy sources. In addition, innovative solutions are offered
for keeping the power supply system reliable and affordable. The book also discusses the effects
of renewable energy on the cost of electricity per kilowatt-hour. Low costs are very important
since energy is so intertwined with the economy – any increase in the price of electricity has a
significant impact on the cost of products and services.
Power systems are not based on feelings and opinions, but on scientific and technical facts.
Therefore, some mathematics and physics are presented in this book. Nevertheless, readers
without a technical background should also be able to understand the issues displayed.
Writing this book took much more effort than initially expected. Large levels of intermit-
tent power sources in a system have an impact on the requirements for contingency reserves,
on balancing electricity production and demand, on supplying reactive power and on costs.
This required searching for actual data on the output of solar panels and wind turbines, in
combination with actual power demand patterns. Fortunately, most European and USA-based
transmission system operators make the necessary information available on the internet,
although often heavy number crunching was required to transform this data into a workable
format. Also, the International Energy Agency always provides useful information. Yet, this
book does not pretend to cover every aspect of the issues at stake. Hopefully it helps the reader
to gain more insight into the matter and serve towards achieving the best solutions.
I am indebted to Wärtsilä for offering the possibility to write this book. The continuous
support from a number of co-workers during its preparation is highly appreciated. In alpha-
betical order, I would particularly like to mention Christian Hultholm, Jaime Lopez, Jiipee
Mattila, Jussi Laitinen, Kärt Aavik, Kenneth Engblom, Kimi Arima, Mats Östman, Niklas
Wägar and Svante Bethlehem. I am especially grateful to my wife Anna Martha who allowed
me to dedicate so much time to writing of this book.
1 How to secure the
electricity supply in a
changing world
The economy is largely built on a reliable supply of cheap electricity. A challenge
is to keep the supply system stable and affordable with the rapid expansion of
intermittent renewable energy sources. The new system cannot just be built on
top of the old one. To make the integration successful and to ensure prosperity in
the future, new technical solutions and market conditions are needed. Business
as usual is not an option for the power sector.
12  Power supply challenges

1.1. An affordable, reliable and sustainable supply of electricity


Electricity is all around us. Without electricity, communications, industrial activities and
services come to a halt. Households suffer badly when the power supply stops. Thanks
to electricity, life in hot regions is bearable. Agricultural products can be treated and
stored for extended periods of time with electrical chilling. Soon, electric vehicles will
be common on the roads. Before long, electricity will be the major energy carrier for
energy consumers.
Electricity production is still mostly based on fossil fuels. Because of the emis-
sions that these fuels produce during combustion, legislators and society in general
are demanding more renewable energy sources. However, the advent of renewable
energy based on, for example, solar radiation and wind is creating challenges in
maintaining the delicate balance between electricity production and demand. Power
plants charged with the balancing task have to adapt their output faster and more
frequently than before. Errors in forecasting electricity production from renewable
sources add up to errors in demand prediction. Consequently, more reserve capacity,
with a much more responsive character than in the past, is needed. Traditional
steam-based power plants lack this flexibility. However, agile generating techniques
exist that have the ability to assist in accommodating vast amounts of renewable
electricity sources and help, in so doing, reducing use of fossil fuels.
According to the media, energy storage, smart grids, huge transcontinental
power transmission lines and demand side management can solve the issues arising
from the intermittency of renewable electricity sources. Such news should be judged
with great care.
It would be convenient to have affordable storage systems playing a major role
in balancing electricity supply with demand. Storing electrical energy directly as
electricity is not yet possible in practice. Energy has to be stored chemically as in
fuel and in batteries, or mechanically as in flywheels, compressed air and raised
water-reservoir levels. Heat from concentrated solar power systems can be tem-
porarily stored in molten salts for steam generation at a later point in time. The
challenge is to find economic storage systems that have the right properties to serve
the balancing of electricity generation and demand. The required storage proper-
ties depend on the type of balancing required. Flywheels might help for short-term
frequency regulation in time spans of a few seconds, while batteries can help to
cover unbalances up to an hour and pumped hydro can take care of smoothing in
24-hour intervals. However, no storage systems exist yet that can substitute the
use of fuels such as natural gas and coal in covering, for example, seasonal lacks in
power output from renewable sources, such as those occurring with solar PV output
during the darker seasons.
High winds can occur in continent-wide areas, so smoothing wind-turbine
output with long transmission lines is not an effective option. Using excess electricity
during the peak output periods of solar panels and wind turbines for water heating
and chilling is a better option. Smart appliances and smart meters in households
1.  How to secure electricity supply in a changing worldy  13

Gross Domestic Product (PPP) per capita per year in 2009


(year 2000 US$)
40 000
North America
35 000

30 000

25 000
Europe
20 000

15 000
World average
10 000 China
Latin America
Middle East
5 000
Asia (ex China)
Africa
0
0 2 000 4 000 6 000 8 000 10 000 12 000 14 000
Electricity use per capita per year (kWh)

Figure 1.1.  There is a direct relationship between the amount of electrical energy used
(kWh) and wealth levels, as expressed in gross domestic product based on purchasing
power parity (PPP).

appear to offer only very limited possibilities for balancing the supply of electricity
with the demand.
The use of electrical energy is directly linked with economic value, as can be
seen in figure 1.1 In contrast to common belief, the domestic use of electricity in
households is, on a global average, less than a quarter of the total electricity use. The
large remaining portion is consumed by industrial users and by commercial users to
create economic value. Electricity is, therefore, primarily a value creator.
A number of conclusions can be drawn from the relationship between gross
domestic product and electricity use as shown in figure 1.1 Simply said, if the power
supply in Africa would increase by a factor of five, the economy might potentially
also grow by a factor of five and much poverty would disappear. In addition, if North
America would lower the intensity of electricity used in its economy to the European
level, electricity consumption might be lowered by some 20% without losing any
wealth. The use of more efficient appliances, better building insulation, and a large-
scale introduction of LED lighting are expected to contribute to reduced electricity
use in the USA. In Europe, by contrast, the replacing of gas-fuelled heating with
electric heat pumps and the advent of electric vehicles might lead to some increase
in electricity use. China appears to closely follow the global trend line between elec-
tricity use and GDP. The Middle East is clearly an outlier: cheap fuel, hot climate,
and relatively low industrial output result in low GDP creation per unit of electric
14  Power supply challenges

energy. Nevertheless, although economic boundary condi-


tions differ from country to country, electricity use and
economic welfare are closely related.
Despite its excellent value to society, electricity has
to be affordable. High electricity prices can be a reason 50 watt
for energy-intensive industries to move to a country with
lower prices. Distinguishing between the cost, price and
value of an economic commodity such as electrical energy, Factor 40 higher pro-
helps in better understanding the mechanisms leading to ductivity with elec-
tricity
affordability.
The basic costs of electricity consist of the cost of
the capital investment for the generating unit, the cost of
electricity transportation and distribution facilities, the
cost of fuel and the cost for operation and maintenance.
The price customers pay for electrical energy normally
contains at least these basic costs, with profit margins and 2000 watt
government-imposed taxes added. The ultimate economic
value for the customer per kilowatt-hour delivered should
naturally be higher than the price that the customer pays.
A domestic consumer generally pays more per kWh Figure 1.2.  An example of
than an industrial user. This is partly because of higher where electricity substantially
distribution and retail costs, but also because profit mar- increases productivity.
gins and levies are generally higher in the case of private
customers. For an aluminium smelter, the value, price, and cost of electricity are
basically close together since energy costs heavily determine the end-product costs.
For a scientist, banker, or family member using a desktop computer, the cost, price
and especially the value of electricity can be factors different. Computers raise pro-
ductivity so much that the price of the electricity to run them is almost irrelevant

Cost, price and value of electricity Cost, price and value of electricity
for an aluminium smelter for a household
120 120
100 100
Eurocents per kWh

Eurocents per kWh

80 80
60 60
40 40
20 20
0 0
Cost Price Value Cost Price Value

Figure 1.3.  The cost, price and value of electricity compared


1.  How to secure electricity supply in a changing worldy  15

Figure 1.4.  The densely populated and polluted environment were created in the new
industrial cities during the Industrial Revolution (1760–1840).

for the user. In households, a 2 kW vacuum cleaner has the same power as 40 people
using dustpans and brushes. One hour of vacuum cleaning might cost 0.50 € for the
electricity, but hiring 40 cleaning people instead might cost at least 500 € in wealthy
economies.
The social costs of electricity can cause the real costs to be higher than the sum
of the costs for capital, fuel and operations plus maintenance. Such social costs
include, among other things, the value of the environmental damage caused as a
result of pollution from the fuel production and from the emissions. Subsidies for
mining jobs also have to be included in the social costs. Politicians might claim the
creation of a substantial number of jobs connected with the introduction of renew-
able energy, but such jobs can also be seen, at least partly, as social costs as long as
subsidies dominate the market for renewables. Ultimately, the integral economic
value of a product such as electricity should at least exceed all the costs of making
that product. If the cost of electricity exceeds its value, using electricity will be a
luxury and a burden on the economy, without creating wealth.
Nevertheless, the acceptability of neglecting social costs depends to a large
extent on the actual wealth level of the particular country. When people are starving,
items such as food and water are urgently needed, and in such cases some connected
16  Power supply challenges

Pure sine wave Sine wave distorted with 3rd and


5th harmonics
1 1
Relative voltage

Relative voltage
0.5 0.5
0 0
–0.5 –0.5
–1 –1
0 10 20 30 0 10 20 30
Time (ms) Time (ms)

Figure 1.5.  A clean 50 Hz sine wave and a sine wave distorted with harmonics.

environmental damage is just taken for granted. The industrial revolution in the 18th
and 19th centuries had destructive effects on the environment, but the resulting
increase in the level of prosperity ultimately released money for repairs and improve-
ments. Enforcing the same environmental standards globally for electricity produc-
tion, regardless of whether it is in emerging economies or in the affluent areas of the
world is, therefore, not fair if the associated costs are high.
While many people use power as a synonym for electrical energy, this book
will distinguish between power and energy. It is scientifically incorrect to state
that a machine or a power plant can produce power, since power is the capacity to
deliver energy. A car can have an engine with a maximum power capacity of 125 kW
(kilowatts), but as long as the engine is not running, no energy is sent to the wheels.
Driving the car for one hour at full power means that the engine delivers an amount
of energy equalling 125 kW ∙ 1 h = 125 kWh (kilowatt-hours). An electric power sta-
tion of 360 MW (megawatts) constantly running at full output during 4380 hours,
equalling half a year, produces 360 ∙ 4380 = 1576800 MWh (megawatt-hours), or
almost 1.6 TWh (terawatt-hours) of electrical energy.
High reliability in supplying quality electricity is obviously important to energy
consumers, who generally require that their need for electric energy is fulfilled at
any time. Failure to supply electricity will at least be a nuisance, and generally also
results in financial losses. Users in commercial and industrial environments expect
a power supply system reliability of at least 99.99%, meaning that the supply fails on
average in total only 53 minutes per year. For applications where a constant avail-
ability of electricity is crucial, uninterruptable power supply systems and backup
generators are common practice. For some applications, such as data centres and
hospital operating theatres, a supply reliability of over 99.999% is required.
A reliable supply of electricity also requires that the voltage and frequency are
maintained within narrow limits. In addition, the delivered voltage should be clean
and not excessively superseded by harmonic or random distortions, i.e. voltage varia-
tions with a frequency other than the basic frequency of 50 Hz or 60 Hz. Distor-
tions are caused by control electronics and by lighting systems such as LEDs. If the
1.  How to secure electricity supply in a changing worldy  17

voltage deviates too much in value and shape from the standards, the performance of
the users’ equipment will be detrimentally affected, and the equipment might even
be damaged. Quality electricity means high supply reliability of the proper voltage.
The electricity supply should also be sustainable. The burden imposed on the
environment should be acceptable, while natural resources have to be used as effi-
ciently as possible. Technologies are available nowadays to achieve very low emis-
sions of pollutants, including nitrogen oxides (NOX) and sulphur oxides (SOX). Both
have negative impact on air quality, and cause acidification of water basins and soil.
As an example of emission reductions, the power sector in the USA was respon-
sible for 6.2 Mtonnes of NOX in 1995, but for only 2.2 Mtonnes in 2009, thanks to
exhaust gas cleaning and cleaner fuels. Yet, total fossil fuel consumption in the USA,
meaning oil, gas and coal together, was roughly the same in 2009 as it was in 1995.
Another issue is global warming. Globally, the power sector is responsible for
roughly a quarter of anthropogenic CO2 emissions. The European Union aims to
reduce greenhouse gas emissions by some 85% from the 1990 levels by the year
2050. The power sector should be emitting zero greenhouse gases by that time. To
achieve this, a reduction in energy consumption, the large-scale introduction of
renewable energy sources, and carbon capture and sequestration (CCS) for fossil fuel
applications are seen as being the major measures. In this context, it is important to
know that power plants are long-term investments with a technical life exceeding 40
years. The EU policy means that newly built power plants that are not prepared for
CCS might face early retirement.
However, an excessively abrupt weaning from fossil fuel usage in order to bring
down CO2 emissions will disrupt the economy. The reason behind this is that per
unit of delivered energy most renewable energy sources are more expensive than

30
NOX and SO2 concentrations in the air

SO2
25

20
(ppb)

15
NOX
10

0
1980 1990 2000 2010
Year

Figure 1.6.  The substantial decline in average concentrations of NOX and SO2 in the
USA’s ambient air (source EPA).
18  Power supply challenges

fossil fuels. Furthermore, energy sources based on wind, solar radiation, tidal flows,
and wave energy are by nature variable in output.
The International Energy Agency (IEA) has estimated that just 3.7 %, or
0.8 PWh, of the total global electrical energy demand was derived from renewable
sources in 2010, excluding hydropower. A large part of this is based on biomass, pri-
marily wood. Wood is often used in existing coal-fired power plants via co-firing or
supplementary firing. Burning wood in power stations is heavily subsidized in some
countries, but the positive effect on reducing greenhouse gas emissions is question-
able. Estimates are that forestry activities and the transportation costs involved
might already result in 200 g/kWh in CO2 emissions, i.e. almost the same amount
of CO2 that a natural-gas-fired cogeneration plant emits. If hydropower is included
in the renewables, some 19.7 % of electricity is currently derived from renewable
sources.
The effort required to increase the amount of renewable energy is huge. Inevi-
tably, fossil-based power plants will still be needed for many decades. In any case,
fully abstaining from the use of fossil fuels is difficult, since these energy sources
can easily be stored in large quantities. In particular, natural gas can serve as a ver-
satile, cheap and relatively low-carbon backup battery for balancing the intermittent
electricity supply coming from wind, solar radiation and tidal-flow generators. Nev-
ertheless, fossil fuel resources are ultimately finite. Expectations are that the global
demand for electrical energy will almost double over the coming 20 years. Therefore,
maximum fuel efficiency is required and any wasting and flaring of fuels should be
avoided. The goal should ultimately be to achieve a gradual shift to affordable renew-
able energy sources with mature equipment having sufficient warranties from reliable
manufacturers.

1.2. Challenges of renewable energy sources.


Electricity demand has always shown variability. Short-term variations in demand occur
because electricity consumers switch their appliances on and off at random. The net
effect of this on demand is small and conventional generators can adapt their output
accordingly. Moreover, there are daily patterns caused by typical societal behaviour,
where people go to work or school in the morning and return home in the evening, and
finally go to bed. These daily patterns are affected by the seasons, since in the colder
regions more lighting and heat are required in the wintertime. In hot regions, air condi-
tioning is needed in the summer.
In many areas, seasonal patterns can clearly be distinguished. Figure 1.7 has
17.520 data points to illustrate how the power demand varies in the north-western
part of Germany during a full year. Each weekend, there is a sharp drop in demand.
The variability in output of fuel-based power plants is much higher even than the
variability in demand shown in figure 1.7 This is because of the intermittent output
of a substantial amount of renewable electricity sources in the system. It is notable
that at the end of the year, when the labour force stops work for the Christmas
1.  How to secure electricity supply in a changing worldy  19

Power supply data Tennet region, Germany, year 2012

15 000
Power supply (MW)

10 000

5 000

Figure 1.7.  An example of the dynamic pattern in the electricity demand in the Tennet
region of Germany (data from Tennet).

holidays, the demand is very low. During this holiday period, strong winds were
prevalent over Germany, resulting in excess electricity that had to be exported to
neighbouring countries for a negative fee of up to 200 €/MWh.
Electricity generators based on renewable energy sources such as wind, sunshine
and tidal flows are generally granted unrestricted feed-in into the electricity grid.
Their output however depends heavily on the weather and the time of day. More-
over, their output is never fully predictable and is sometimes even close to zero.
Figure 1.8 illustrates the variability in output of wind turbines and solar PV panels
in the German 50Hertz TSO (Transmission System Operator) region during week

Week 26, 2012, 50 Hertz TSO are, Germany


125 000

Others
100 000
Power (MW)

75 000
Wind
5 000
Solar
2 500

0
Mon Tue Wed Thu Fri Sat Sun

Figure 1.8.  Wind and solar -based power output, and the remaining supply from other
sources in the German 50Hertz TSO region, week 26, 2012 (cumulative curve, the black
arrows give the extremes for ‘others’).
20  Power supply challenges

26, 2012. Early on Monday, wind turbines generated almost all the power that was
needed. On Thursday morning at 8 am, however, the output from wind and solar
sources was so low that 10.4 GW had to be derived from other sources.
Since electricity transmission and distribution grids have virtually no energy
storage capacity, the production and consumption of electricity have to be precisely
matched. If the driving power for the generators exceeds the electricity consumption
+ system losses, the generators will increase their speed and, simultaneously, the
frequency in the system will go up. Alternatively, if demand is higher than supply,
the frequency will drop. For the system to operate properly, and for many sensitive
applications, the frequency has to remain within narrow limits. Therefore, generators
are equipped with controllers that can correct their output depending upon the devi-
ation from the desired system frequency. Variation in power plant output is therefore
necessary, but it is not economic to run a power plant consisting of a single gener-
ating unit in a wide load range. Technical restrictions also limit a single generator
from having a wide output range. At low loads, the fuel efficiency is low while the
maintenance costs per kWh are high. Therefore, generators are switched off if their
load is below a certain threshold. Conversely, if the generators that are online cannot
meet an increase in demand, additional generators have to be switched on.
With much intermittent renewable capacity in the system, the balancing task of
the fuel-based and hydro-based generators is rapidly increasing. When the sun sets,
the output from photo-voltaic cells (PV) drops to zero, while electricity demand gen-
erally increases. This results in the dispatchable generating capacity having to ramp
up its output significantly. Figure 1.9 gives an example of the rapidly changing output
from all the wind turbines in the German Amprion TSO region during the 24 hours
of April 28, 2012. This illustrates a typical example of the passage of a depression.
Two substantial increases in the wind-power output of up to 1 GW per hour were

2 500
Power output wind turbines

2000

1 500
1.2 GW/hour
(MW)

decline in output
1 000

500

0
0 6 12 18 24
Hours of April 28, 2012

Figure 1.9.  Large differences in power output from wind turbines in the German Amp-
rion TSO region on April 28, 2012.
1.  How to secure electricity supply in a changing worldy  21

observed. The large decline in wind-power output, from 1.8 GW to 0.6 GW, in the
time span from 11 am to 12 am required a large amount of fast backing-up by power
plants. Even worse situations occur when the wind reaches gale force and wind tur-
bines have to be stopped in order to avoid physical damage. Because of this, backup
power plants have to be increasingly flexible.
Transmission system operators (TSOs) try to introduce Demand Side Manage-
ment (DSM) for balancing. Sometimes, it is called Demand System Response. Typical
electric appliances, such as refrigerators and air-conditioners, can be switched off for
a while. The use of washing machines and laundry dryers can often be postponed to
when the general demand for electricity is dropping. This requires smart appliances
that respond to a signal from the grid operator. Variable pricing of electricity might
also help, and for this smart meters with a momentary tariff indicator are needed. Figure
Nevertheless, such demand management measures can only be part of the 1.10.  The
typical elec-
solution to keep the system stable. A huge number of appliances would have to be
tric power of a
controlled to have any noticeable effect. As an example, using DSM to compensate laundry dryer is
for the 1.2 GW decrease in wind turbine output, as shown in Figure 1.9, requires 2.7 kW.
switching off the equivalent of 450000 laundry dryers. A laundry dryer runs on
average for some 100 hours per year. The probability that a laundry dryer is run-
ning at any particular time is, therefore, only 1%. To sum up, it is not expected that
domestic electricity demand can be shifted by more than a few percent through
the use of smart appliances and smart meters. Better DSM opportunities might be
present with industrial users of electricity.

1.3. Faults and failures in electricity supply systems


In addition to the need for normal balancing, faults and failures can and do occur in
electricity supply systems. These malfunctions are also called contingencies, and they
affect the balance between supply and demand. A failing power plant results in an
instantaneous loss of electricity supply. Immediately upon the occurrence of such a loss
in generating capacity, the rotating inertia in the system helps to avoid an abrupt change
in frequency. A consequent drop in frequency is unavoidable, but system operators allo-
cate spare output from the generators that are online to compensate for the lost unit.
This spare output is called primary reserve. After application of the primary reserves,
additional generating capacity is rapidly activated to restore the frequency to the desired
value so that the primary reserves are available again for the next contingency. This addi-
tional capacity is called secondary reserve. With a growing fraction of the power capacity
derived from non-dispatchable generators in the system, it becomes increasingly diffi-
cult to have adequate backup capacity within the system. Having just a few large power
plants online is very risky, since then the relative effect of one power plant failing on the
dispatchable capacity is large. Modern contingency reserves have to consist of smaller
agile power plants that are well distributed across the area to be served.
Cogeneration units are an example of such distributed power plants. Cogenera-
tion of heat and power (CHP) is an effective means of improving fuel efficiency and
22  Power supply challenges

reducing greenhouse gas emissions. With an adequate number of such local genera-
tors in a system, these units can also be utilised for frequency control and balancing.
Again, highly flexible power plants with fast ramping rates and short starting and
stopping times will be needed for balancing electricity production and demand.
Faults in the transmission and distribution system cannot be avoided. Trees may
fall on high voltage lines, and icy rain in combination with high winds can damage
the wires. Excavations frequently cause damage to underground cables. Decentral-
ized generation is beneficial in this respect, since it reduces the dependence on a few
distant generators and long power lines. Because of their increased contribution to
electric generating capacity, decentralized generators should also be able to comply
with the grid codes set by transmission and distribution system operators for large
power plants. Being able to ride through a short circuit is an example one of the new
requirements for local generators (see appendix 2).

1.4. Balancing supply and demand in energy markets


W hen electricity was supplied only by fully integrated utilities, all costs in the system
were supposed to be covered by the tariff charged to the customers. In such a system,
any profit goes to the system owner, which is often the state, a province or a munici-
pality. The system owners are also responsible for any financial losses. In some countries,
electricity is even subsidised. Planning expansions to, and renewals of, the generation

Transformer
drum

Power
poles

Power
substation
Transmission
substation Transformer
Power
plant

High voltage
transmission lines

Figure 1.11.  An integrated power company producing and delivering electricity to end
users.
1.  How to secure electricity supply in a changing worldy  23

Emission limits, Politicians


emission tax and Energy tax and renewable
renewable subsidies subsidies
Government

Competing Large industrial


generating Regulator customers
company 1

Transmission
system Large industrial
Competing
operator customers with self
generating
generation
company 2

Distribution system
Subsidised Small industrial
operator
intermittent customers
renewable generation

Commercials
Competing Competing
electricity retailer A electricity retailer B
Households

Competing
electricity retailer C

Variable rules Variable emission limits, Variable electricity flow


tax and subsidies

Figure 1.12.  An unbundled power sector affected by many players.

and distribution system is easy in such circumstances. For this reason some govern-
ments and politicians prefer a situation whereby the electricity supply is fully controlled
by integrated utilities, with perhaps a few independent electricity producers feeding into
the grid.
However, the free-market thinking at the end of the twentieth century advocated
economic liberalisation, with privatisation and deregulation in all segments of all
markets. By having private investors take over the role of the public sector, produc-
tivity was supposed to increase and the costs to consumers would then be reduced.
A power supply run by the public sector was, and often is, considered to be
bureaucratic, ineffective and less customer friendly. Is this truly the case? That is the
big question. Currently, the liberalised and unbundled power sector complains of
permanent interference from policy makers with ever-changing rules and high sub-
sidies for some types of generation, making it difficult to invest in new power plants.
Yet, extensive lobbying continues simultaneously by the different stakeholders for
24  Power supply challenges

getting preferential rules for their typical facility or technology. Constant changing
of the rules creates difficulties for long-term investments.
Power stations have a very long technical life, and transmission and distribution
lines last even longer. This is the reason that some countries, especially in Asia, have
decided not to adopt the liberalised market model, which is generally dominated
by short-term profit making and quarterly results. Moreover, in many countries
with a liberalised electricity market, the government still derives much income by
taxing energy use and by charging value added tax. The ultimate price of electricity
to domestic consumers has, in general, not decreased as a result of the new open
markets.
Grid frequency, voltage levels and reliability all have to be guaranteed, even in
an open electricity market. Therefore, independent transmission system operators
(TSOs) are charged with the control of frequency and voltage, and with setting
rules for maintaining grid stability and supply reliability. TSOs estimate the power
needs for the near future with elaborate prediction models, and use a market
mechanism to ensure that sufficient generating capacity will be available. With
the introduction of much intermittent generating capacity from renewables, the
uncertainty in predicting the output required from non-renewable power plants is
growing. As an example, the large changes in output from wind turbines as shown
in Figure 1.9, were not predicted by the forecasting models, as can be seen from
figure 1.13.
In the simplest open market approach, a power plant is remunerated only for the
energy delivered. The producer that offers the cheapest electricity would be first in
the merit order in a national or regional energy market. In this simple market model,
the TSO requires electricity producers to include all relevant services, including

2 500
Power output wind turbines

2 000

Actual
1 500
(MW)

1 000
Forecast

500

0
0 6 12 18 24
Hours of April 28, 2012

Figure 1.13.  An example of a large deviation between the predicted and the actual
power output from wind turbines in the German Amprion TSO region, April 28, 2012.
1.  How to secure electricity supply in a changing worldy  25

1
Imbalance (GW)

0.5

–0.5

–1
0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 24:00
Time

50.10
50.08
50.06
50.04
50.02
ƒ (Hz)

50.00
49.98
49.96
June 2003
49.94
June 2006
49.92
49.90
01 0:00
02 0:00
03 0:00
04 0:00
05 0:00
06 0:00
07 0:00
08 0:00
09 0:00
10 0:00
11 0:00
12 0:00
13 0:00
14 0:00
15 0:00
16 0:00
17 0:00
18 0:00
19 0:00
20 0:00
21 0:00
22 0:00
23 0:00
00
0:
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
:0
00

The 24 hours of April 10, 2013


50.15
50.10
Grid frequency (Hz)

50.05
50.00
49.95
49.90
49.85

Figure 1.14.  Examples of imbalances caused by electricity trading (data from KEMA
report 74100846-ETD/SDA 12-00079, Swissgrid and TENNET.
26  Power supply challenges

backup for failing power plants and frequency control in their energy delivery
offering.
In a more extended market model, power plants can be remunerated for the avail-
ability of reserve power and for their capability to achieve fast ramping up or down
of their output. Even factors such as starting up reliability and supply reliability,
might be worth rewarding. Manufacturers of energy storage technologies are aiming
for financial compensation for the balancing capabilities of their products. Apart
from pumped-hydro storage, most technologies for short and medium-term storage
are still under development, and researchers are eager to promote their technologies
to subsidy providers.
Electricity supply markets generally operate by offering energy in fixed time
spans, such as in hourly or even 15 minute intervals. This approach gives rise to
periodic deviations in grid frequency. This is illustrated in figure 1. 14. Frequency
stability has, therefore, decreased since the introduction of open electricity markets.
Each time a trading time span ends, or begins, power plants increase or decrease
their power output. This has to be compensated for by the frequency regulation
capacity of the power plants, which was originally intended for occasional contingen-
cies, such as the loss of a power plant. Sluggishly reacting power plants have diffi-
culty in restoring the grid frequency to within its required range.
In chapter 7, this book will show how a properly selected generating portfolio
in an electricity supply system can improve system stability with reduced costs and
higher reliability. With a proper approach, this stability can even be reached with a
high proportion of intermittent renewable generation in the system. Much intermit-
tent generation inevitably reduces the utilisation factor of the other power plants.
The consequence of a low utilisation factor is higher specific capital costs (€/MWh)
for fuel-based power plants. Low investment costs will, therefore, be a key element
for new power generating capacity.
The shift towards more renewable generation in the system will certainly reduce
fossil fuel consumption. However, the consequent decrease in the utilisation factor of
the other power plants will inevitably increase the capital costs per kWh produced.
Moreover, fuel-based power plants will have to be far more flexible in the future,
with frequent starts and stops and high ramping rates in output.

1.5. Conclusions
A steady growth in electricity use, coinciding with concerns for sustainability, creates
substantial challenges. Policy makers interfere increasingly with markets and use subsi-
dies and levies to achieve their targets. Investors face uncertainty of profitability because
of frequently changing boundary conditions. Dispatchers of power supply systems
have to live with the challenges of variable outputs of renewable energy sources and
the uncertainties from forecasting errors. To compensate for the unpredictability of the
markets and to backup the intermittent output of renewables, a new level of flexibility is
needed in power systems.
2 Balancing the
electricity supply in
case of calamities
Stability in electricity supply systems has to be maintained even during disturbances
such as a major short circuit, generator failure or losing a large load. In keeping the
system stable, the role of rotating inertia is essential. When integrating renewables
with no or low inertia to the system, the balancing becomes more difficult. To
avoid risks of frequency collapses and blackouts, new solutions are needed for the
fuel-based backup generation.
30  Power supply challenges

2.1. Matching electricity supply and demand


The main task of an electric power supply operator is to continuously match electricity
generation with electricity demand. Continuous matching is needed since the supply
system as such cannot store electrical energy. If electricity demand systematically
exceeds the power delivered by the machines that drive the generators, the generating
units will respond by decreasing their rotational speed. Consequently, the grid frequency
will drop and the system will collapse in a matter of seconds, resulting in a blackout.
Fortunately, blackouts won’t occur if the unbalance in demand and supply is short-term,
since generating units and electric motor drives have energy stored in their rotating
mass, the so-called inertia. That buffer limits the rate of change in frequency in the
case of an unbalance between generation and demand. The energy stored in this inertia
creates time for the engines or turbines that drive the generators
to adjust their output in order to restore balance.
The stability of the frequency in alternating current sys-
tems is a good measure of balance. Frequency is by definition
Balance
the number of times that a full sine wave occurs per second
in the grid. The international unit denoting frequency is
hertz (Hz). The rotational speed of the generators determines Generation
Inertia
Demand

this frequency, and the so-called nominal value of the fre-


quency depends on the global location. America and Japan
operate using 60 Hz, while most other areas of the world have
50 Hz. In reality, the grid frequency varies somewhat around
the desired value. Figure 2.2 shows the grid frequency in The
Netherlands during a short time span of 5 minutes after 5.00
am on April 12, 2013. Customer demand is never fully con-
stant and generators also have some variability in their output. Figure 2.1.  An illustration of the deli-
Yet, on average the frequency has to match the desired value. cate balance between electricity demand
If the actual frequency deviates for just a small fraction and production, with rotating inertia as a
buffer with some energy stored.
from the desired value, no action is taken to change the
output setting of the generators. There are a number of reasons
for that. Each measurement system is afflicted with some inaccuracy, while control
systems also have some insensitivity. Endeavours to keep the frequency within very
narrow limits, meaning real isochronous operation, would result in overactive control
of the machines that drive the generators, which in turn would lead to unnecessary
wear. Figure 2.2 shows an example of a system with a permitted measurement error
range of +/– 10 mHz, with an additional zone of +/– 10 mHz where no action from
the generator is required. Consequently, the result is a total dead band of +/– 20
mHz. Nevertheless, the average frequency over a prolonged time span should be
exactly 50.000 Hz, and grid operators take action if the cumulative deviation from
this desired value becomes excessive.
In Figure 2.2, the grid frequency exceeds the dead band at 20 seconds after 5.00
am on April 10, 2013. At that moment, the so-called primary control reserve power
2.  Balancing the electricity supply in case of calamities  31

50.05

50.04
Frequency regulation action
50.03
Grid frequency (Hz)

50.02
Additional dead zone
50.01
Allowed
measurement error
50.00

49.99 Additional dead zone

49.98

49.97
0 60 120 180 240 300
Seconds after 5:00 am on April 10, 2013

Figure 2.2.  Example of the rules for frequency control, with an example of actual
frequency variations (frequency data source: TENNET)

plants start to automatically slightly reduce their output settings. Some one hundred
seconds later, the frequency is back within its allowed limits and the control action
of the generators ceases. This frequency regulation action is handled automatically
by the so-called primary reserves. Automatic action is the only option because of the
fast response required to keep the frequency within its narrow band.
The primary control reserves are also known as frequency containment reserves
(FCR). Substantial changes in frequency will occur if a large customer disconnects,
or if a large power station suddenly fails. In modern power supply systems, many
generators are interconnected via the transmission grid. The number of online gener-
ators should be sufficient to ensure that a failure of the largest unit can be absorbed,
to a large extent, by the spare capacity of the other generating units.

2.2. Primary control reserves compensating for the failure of a power plant
It is interesting to analyse what happens when a power plant in an electricity supply
system fails. To simplify such an analysis, we presume a supply system with ten power
plants of the same power capacity. Each of the ten power plants has a nominal power
capacity of 500 MW, and they are all running at 90% of their capacity to provide 10% of
primary control reserves. That would, at least in theory, be sufficient to compensate for
a failure of one of the ten power plants. Nominal power means the nameplate power of
the generating unit, while nominal speed means the generating unit’s normal amount of
revolutions per minute. In this example, the electricity demand that the ten power plants
32  Power supply challenges

1 2 3 4 5 6 7 8 9 10


Transmission lines

Demand by customers

Figure 2.3.  Ten power plants initially supplying the required electricity demand when
one of them, number 7, fails.

supply amounts to 90% of 10 · 500 MW = 4500 MW. It will now be shown what hap-
pens if, for example, power plant number 7 suddenly fails and the system immediately
lacks 450 MW of the required power supply. This example may appear to be somewhat
exaggerated since a sudden loss of 10% of the dispatchable generation is not common.
However, with much renewable capacity in a system, such occurrences are becoming
increasingly realistic. In addition, the effects of unbalance in a system can be clearly
shown with this example.
After the failure of one plant, the nine remaining power plants cannot instanta-
neously ramp up the power output of the machines that drive the generators from the
initial 450 MW to the newly required 500 MW in order to supply the total system
demand of 4500 MW. Power plants need some time to react to a newly desired
output value. Therefore, if no energy was available from the rotating mass (the rota-
tional inertia) in the system, the unbalance would immediately stop all generators
with a resulting blackout. The amount of energy stored in a rotating generating set

Rotational energy ω=2 π f = 2 π n/60


E r=½ l r ω 2

Energy in Energy out

Figure 2.4.  The energy stored within the power supply system as a result of the fly-
wheel effect of the spinning generating units. (f = rotational speed in revolutions per
second, Ir = moment of inertia).
2.  Balancing the electricity supply in case of calamities  33

is linearly proportional with the moment of inertia I, and the square of the running
speed n. Inertia is a property characterising the flywheel effect of the rotating mass.
The running speed n gives the number of revolutions per minute of the generator
rotor. In a 50 Hz system, a generator with a single pole pair runs at 3000 rpm. In
this case, the frequency f equals n/60.
The amount of energy Er stored in the rotating inertia of a generating set is gen-
erally expressed as a fraction of the nominal power capacity Pnominal of that generating
set. This fraction is called the inertia constant τI of a generating set:
Er ½.Ir .ω2 Ir
τI= = =½. (2πf)2 . Equation 2.1
Pnominal Pnominal Pnominal
The dimension τI of the inertia constant is the same as that of time, and is
expressed in joule/watt = J/(J/s) = s (second). The inertia constant of a large gen-
erating unit lies in the range between 5 and 10 s. This means that when a 500 MW
generating unit is running at its nominal speed, its rotating parts have 2500 MJ
of rotational energy in the case of a τI of 5 s, and 5000 MJ for a τI of 10 s. This is
also the amount of energy that has to be transferred to the rotating parts when the
generating unit is started up and accelerated to its nominal speed. If the 5000 MJ of
energy during this acceleration is supplied to the rotating inertia with a machine that
uses natural gas for fuel, we can calculate the amount of gas required. If the fuel effi-
ciency of the driving machine is 40% and the natural gas has a lower heating value
of 36 MJ/m3, it requires 5000/36 · 100/40) · 350 m3 of gas to bring the rotor up to
nominal speed. Such an amount of gas provides enough energy to heat up 35000
litres of water from 20 °C to 100 °C, for preparing 280000 cups of tea. A cluster of
3000 car batteries can also deliver the required amount of energy. This illustrates
that although the energy stored in the rotating inertia of a generating set is not elec-
trical, it is nevertheless an impressive amount of energy.

2.2.1. The usefulness of the inertia constant τI


The inertia constant τI is very helpful in getting a first impression as to how fast a gen-
erating unit will change speed in case of an unbalance between the power supply to the
generator and power demand. Unbalance occurs when the electrical load changes while
the power supplied to the generator shaft from its prime mover, i.e. the driving engine
or turbine, remains the same. For an inertia constant of 5 s, the amount of energy in the
rotating parts is enough to supply the nominal load of the generator for 5 s without any
energy input from its prime mover. After that time span, the generator will come to a
standstill. Furthermore, if the prime mover were to supply 90% of the nominal load to
the generator, while the load of the generator equals the nominal load, the rotational
energy would be enough to cover the unbalance for 50 s. Thus, it takes ten times longer
than in the case of no power supply from the prime mover before standstill is reached.
However, even in the latter case, the frequency will rapidly reach a value outside the
permissible range.
34  Power supply challenges

Equation 2.1 reveals that the rotational energy of the generating set is propor-
tional with the square of the instantaneous running speed. This means that at higher
speeds, there is much more energy in the inertia than at lower speeds. Therefore, the
frequency will not decrease linearly with time if there is a fixed unbalance between
the power supply to the generator and the generator load. Figure 2.5 shows how the
frequency of the grid served by the nine remaining generators (Figure 2.3) decreases
if each generator receives 450 MW from its prime mover while the combined load
remains 4500 MW. Each of the nine running generators should receive 500 MW to
avoid an unbalance. Therefore, each generating unit ‘feels’ a power supply deficit of
50 MW. The frequency curve in Figure 2.5 is a representation of equation 2.2. The
derivation of equation 2.2 requires some considerable mathematical manipulation.
The interested reader can find the derivation of equation 2.2 in Appendix 1.
ΔP f 2nominal
f(t)= √(f 2nominal + . . t) Equation 2.2.
Pnominal τi

Prominal= 500 MW, ΔP= –50 MW, Pinitial =450 MW


60

50
Grid frequency (Hz)

40

30

20

10

0
0 20 40 60 80 100
Time (seconds)

Figure 2.5.  The decline in speed for a generator where the generator load constantly
exceeds its driving shaft power by 50 MW (inertia constant τI = 10 s, constant load pre-
sumed).

The drop in frequency during the very first seconds following a major contin-
gency event is a perfect indicator of the amount of unbalance. In the beginning, the
rotating frequency of the generator set is still very close to the nominal frequency
(50 Hz or 60 Hz) and the approximation can, therefore, be made that the initial drop
in frequency per unit of time df/dt equals:
df ΔP
=
dt 2τi Pnominal f nominal Equation 2.3.
2.  Balancing the electricity supply in case of calamities  35

Equation 2.3 reveals that in our example of a τI of 10 s, a power deficit of 50 MW


per generator with a nominal power of 500 MW at a frequency of 50 Hz, results in
a change in frequency of exactly 0.25 Hz per second at the start of the occurrence.
This is indicated in Figure 2.6 by the brown line. Would the power deficit per gen-
erator have been only 25 MW, the decline in frequency would have been halved to
0.125 Hz/s. This simple relationship between unbalance size and the initial frequency
change is very convenient, especially in island operation where just a few generators
have to maintain grid stability. Based on the value of the inclination in frequency
versus time, the fuel supply to the machine driving the generator can be immediately
and adequately adapted so that no time is lost in restoring the generator frequency.
The horizontal red lines in Figure 2.6 give the maximum allowed dynamic frequency
limits in the Continental Europe synchronous area. The green line in Figure 2.6 rep-
resents the nominal grid frequency of 50.000 Hz.

Frequency decline without control action


51

50.5
Frequency (Hz)

50

49.5

49
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (seconds)

Figure 2.6.  A close-up of the first 4 seconds of figure 2.5; the thick brown line gives
the decline in grid frequency for a 10% unbalance in the system due to the tripping of a
power plant (τI = 10 s).

2.2.2. Self-regulating power in an electricity grid


If the grid frequency decreases, electricity demand automatically goes down slightly.
This is primarily caused by the synchronous electric motors in the grid demanding less
power, since their load declines along with their running speed. This is called the self-
regulating power of the supply system. The power requirement of synchronous motor-
driven pumps can decrease by 6% per Hz frequency drop. For motor-driven applications
with a constant torque, power demand can decrease 2% per Hz. Since synchronous
motors form just a fraction of the total load, the self-regulating power of the system
in industrialised countries is generally presumed to be 1% per Hz deviation from the
36  Power supply challenges

nominal frequency. In areas with minor industrial activities, the self-regulating power of
the grid will be close to zero.
The positive effect of self-regulating power should not be overestimated. If
the grid frequency drops from the desired value of 50 Hz to 49.8 Hz, the self-reg-
ulating power lowers demand by only 0.2 · 1% = 0.2%. This 0.2% is only a 9 MW
reduction from the 4500 MW total load in our ten generator system example. If,
however, the grid frequency would drop from 50 Hz all the way down to 40 Hz,
the decrease in electricity demand because of the self-regulating power would be
10 Hz · 1% / Hz = 10%, i.e. 450 MW in our example. This renders a 50 MW reduc-
tion in demand for each of the nine generators that remained online following the
trip of one machine. Consequently, at a frequency of 40 Hz, generation and demand
are matched again if the nine power plants keep their power output setting at the
initial 450 MW. In reality, it will be difficult for the generators to keep their power
output constant when the frequency decreases so much. The power output of the
prime mover that drives the generator is proportional with the product of torque
M and running speed n. If the grid frequency decreases from 50 Hz to 40 Hz, the
driving torque M has to increase by a factor of 50/40 = 1.25 in order to deliver the
same power to the generator. This can easily result in mechanical overload. At the
same time, turbo machinery in particular is burdened with natural frequencies of the
rotor system that limit its range in running speeds. Generators suffer when running
at low frequencies because of the so-called magnetic over fluxing, which results in
possibly harmful overheating. In practice, the self-regulating power of a grid hardly
offers any help in balancing.

Pnominal per generator 500 MW, ΔP=50 MW, 1%/Hz


self regulation of grid load
60

50
Grid frequency (Hz)

40
Load with self regulation
30

20
Fixed load
10

0
0 25 50 75 100 125 150
Time (seconds)

Figure 2.7.  Mitigation of the decrease in frequency due to self regulation of grid load in
the case of a 50 MW initial unbalance between generator output and load.
2.  Balancing the electricity supply in case of calamities  37

2.2.3. Explanation of the droop function of generator sets


A decrease in grid frequency from 50 Hz to 40 Hz is excessive and unacceptable
for many applications and generators. The maximum deviation from the nominal
frequency during dynamic events, such as the loss of a generator or the loss of major
load is, therefore, set at +/– 800 mHz in the Continental Europe synchronous control
area. If self-regulation of the load alone would be present, this frequency range would
be heavily exceeded in the case of calamities. Therefore, each electricity supply system
has generators offering primary control reserves that are activated automatically if
the grid frequency exceeds its tightly defined limits. The change in output from the
primary control reserves depends on the extent of the deviation in grid frequency
from the nominal frequency. In other words, the desired output of a generator acting
as primary reserve depends on the actual grid frequency; the lower the frequency, the
higher the output of the primary control reserves. This dependence of the output set
point on the grid frequency is generally called the droop sgenerator (Equation 2.4). The
minus sign in equation 2.4 indicates that a decrease ∆f in grid frequency results in an
increase ∆P in the power output from the generator providing primary reserves.
1 –Δf
ΔPgenerator = . .Pgenerator nominal Equation 2.4.
sgenerator f nominal
This is often re-written in terms of the regulating power Pregulating of the
system:
ΔPgenerator = –Pregulating . Δf , Equation 2.5.

in which:
Pgenerator nominal
Pregulating = Equation 2.6.
sgenerator . fnominal
The droop value setting of the control system lies generally within a range of 2 to
8%. A droop sgenerator of 4% means, for instance, that the power output of the gener-
ator increases from 0% to 100% if the frequency decreases by 4%, say from 50 Hz to
48 Hz. The relationship between output change and frequency deviation is linear. In
measurement and control technology language, this is called a proportional action,
indicated with the letter P. It means that the extra power from the primary control
reserves is only present as long as the deviation from the desired nominal frequency
exists.

2.2.4. The function of primary reserves


As explained above, primary reserves are intended to avoid excessive deviations in fre-
quency during a major occurrence affecting the balance in the electricity supply system.
However, the primary reserves only arrest the frequency temporarily. They cannot
return the frequency to its nominal value. Figure 2.8 illustrates how the power output
38  Power supply challenges

from a 500 MW generator running at 250 MW, and acting as a primary reserve, varies
with frequency for three different droop settings. For a grid frequency of exactly 50Hz,
the output of the generator equals 250 MW. When the grid frequency decreases, the
output from the generator will increase linearly along with it. Conversely, if the grid
frequency increases, the output from the generator will decrease by following the droop
line. Figure 2.8 reveals that the lower the droop percentage is, the heftier the reaction of
the generator will be on deviations from the nominal grid frequency.
At first sight, opting for a very low droop value might be the best option to keep
the grid frequency as close as possible to the desired 50 Hz. However, if the droop is
very small, meaning that the gain factor is very high, the primary reserves may react
fiercely to deviations in frequency. This results in extra wear of the machinery, while
increasing the risk of oscillations and system instability. Some traditional power
plants suffer heavily from rapid changes in output. High-temperature steam boilers
can experience cavitation and thermal shock during sudden load changes. Moreover,
all generators acting as primary control reserves do not have the same dynamic prop-
erties in practice. Each unit has its own delay time when reacting to an increase in
the output set point, and each unit will have its own typical ramp up rate. Until now,
a close to stepwise increase in output was not possible. Nevertheless, some modern
generating techniques can react much faster than traditional units. The typical ramp
up rate for primary control reserves in the Continental Europe synchronous area
system is 100% within 30 seconds (Figure 2.9). After a short delay of, say 2 seconds,
the primary control reserves are supposed to increase their output at a fixed rate.

500 MW generator running at 250 MW


500

400
Generator output (MW)

300

200

100

0
48.0 48.5 49.0 49.5 50.0 50.5 51.0 51.5 52.0
Grid frequency (Hz)

4% droop 3% droop 5% droop Frequency limits

Figure 2.8.  Output from a 500 MW generator acting as a primary control reserve, run-
ning at 250 MW at a nominal grid frequency of 50 Hz, with three different droop settings.
2.  Balancing the electricity supply in case of calamities  39

100
Output primary reserves (%)

80

60

40

20

0
0 5 10 15 20 25 30 35
Time after ramping up command (seconds)

Figure 2.9.  Typical ramp-up rate of primary control reserves.

Let us now return to our example of the electricity supply system illustrated in
Figure 2.3, whereby ten identical generators were each carrying a load of 450 MW
when suddenly one generator tripped. Without any action being taken with respect
to the set point of the prime mover that drives each generator, the grid frequency
would drop to 40 Hz, as shown in Figure 2.7. However, if each of the nine remaining
generators is also used partly for primary frequency control, each with a droop of 4%
as depicted in Figure 2.10, the grid frequency will not decrease all the way down to
40 Hz. Due to the droop setting, the set point for full output of each power plant will
be reached already at a grid frequency of 49.8Hz, as shown in Figure 2.10. Figure 2.6
reveals that after the contingency of one failing power plant, this 49.8 Hz will already
be reached in about 0.5 seconds after the trip of power plant number 7. Therefore,
the new set point that asks for full output of the generators can be presumed to be
present almost immediately after the loss of one of the ten generators in the system
of our example.
However, notwithstanding the quick change in their set points, the nine power
plants that use their additional available capacity for primary control reserves will
not immediately reach their full output. If we presume that the output of these
power plants follow the prescribed ramping up as shown in Figure 2.9, the load from
the grid and the power supplied to the generators will equal each other after about
25 seconds (see Figure 2.11). At that point, the grid frequency reaches its minimum.
According to Figure 2.9, the additional 50 MW needed per generator that was
lost when one of the original ten generators tripped is fully available from the pri-
mary reserves after 30 seconds. This 50 MW per generator is slightly more than the
grid load requires for staying balanced at the mentioned minimum in frequency. This
is because of the reduction in load created by the self-regulating power. The excess
40  Power supply challenges

energy delivered is then used to accelerate again the rotating masses, the ‘flywheels’,
in the system. Nevertheless, the nominal 50 Hz frequency cannot be reached with
primary control reserves following a droop line. In our example, as soon as the grid
frequency exceeds 49.8 Hz, the output of the primary reserves once again decreases
(see Figure 2.10). That would create another mismatch between power supply and
load. Therefore, a deviation between the actual frequency and the nominal grid
frequency of slightly less than 0.2 Hz will remain where only primary control
reserves are used to restore balance. This deviation between the frequency ultimately
reached with the help of primary reserves and the desired frequency of 50.000 Hz
is called the quasi steady state deviation. Transmission system operators define the
permissible minimum and maximum of this deviation. Extra capacity, the so-called
secondary control reserve, is needed for providing the extra power in the system so
as to restore the grid frequency to the required narrow band around the nominal
frequency.
As mentioned earlier, this narrow frequency band equals only +/– 20 mHz
around 50 Hz in the Continental Europe synchronous area. The application of sec-
ondary control reserves increases the frequency further, so that the primary reserves
can reduce their output and return to their initial load setting at 50 Hz. In other
words, secondary reserves release the primary reserves from their duty. This enables
primary reserves to be ready for the next major occurrence, such as the sudden loss
of a generator or the losing or receiving of a large load.
The previous example whereby primary control reserves equalled exactly the
amount of lost generating capacity, is obviously an exception. In reality, the power

500 MW generator running at 450 MW


500

400
Generator output (MW)

300

200

100

0
48.0 48.5 49.0 49.5 50.0 50.5 51.0 51.5 52.0
Grid frequency (Hz)

Figure 2.10.  The droop function of a generator having a 500 MW nominal power, again
with a droop setting of 4%, but now running at 450 MW at 50 Hz.
2.  Balancing the electricity supply in case of calamities  41

plants in a system do not all have the same


capacity. Primary reserves should be large Only self regulation
Typical point for load shedding
enough to compensate for at least the loss With primary reserves activated
of the largest power plant in a system. In
practice, power plants not dedicated as 50

Grid frequency (Hz)


primary reserves will also provide some 48
compensating power for restoring the grid 46
frequency to its nominal value. Neverthe-
44
less, our example illustrates the way con-
tingencies are handled in a grid system. 42
Transmission system operators 40
specify the maximum allowed dip in 0 25 50 75 100 125 150
frequency, officially called the minimum Time (seconds)
instantaneous frequency after loss of
generation. In the Continental Europe Figure 2.11.  Primary control reserves prevent the system
synchronous area, this value equals 49.2 from decreasing too much in frequency after a loss in gener-
Hz. If the grid frequency drops below this ating capacity (inertia constant 10 s, 1%/Hz self regulation,
50 MW unbalance for a generator with a nominal capacity of
minimum allowed frequency, load shed- 500 MW).
ding will be used to avoid too deep devia-
tions from the nominal frequency. This means that a group of consumers will have
no access to electricity for a while. The example with the 10 power plants of equal
size, where one of the units trips, shows that losing 10% of generating capacity gives
a frequency dip way below 49.2 Hz. Hence, there should be enough power plants in
a system to ensure that the loss of one power plants does not reduce the online gener-
ating capacity by more than about 3%. In particular, systems having a large fraction
of renewable electricity sources need dispatchable power plants of a limited size, and
consequently more of them than in the case of large power plants only.

2.2.5. The consequences of a lower inertia constant


Should the inertia constant of the combined generators in a system decrease, for example,
due to the introduction of a large amount of renewable energy sources that are indirectly
connected to the grid via frequency converters, the grid frequency can drop to quite low
values during a contingency. The red line in Figure 2.12 shows a deep dip in frequency
if the inertia constant of the system is 5 s instead of 10 s. The other conditions are the
same as in Figure 2.11, with 1%/Hz self regulation, 50 MW unbalance per generator and
a nominal generator capacity of 500 MW.
The 46 Hz minimum in the red curve occurs 18 seconds after the calamity that
tripped one of the 10 power plants in the example. At that point, the output of the
primary reserves has not yet reached its maximum, since that occurs 30 seconds
after the trip. After 18 seconds, the output of the primary reserve per generator is
therefore only 18/30 · 50 MW = 30 MW. However, the self-regulating power of the
grid equals 4 · 1/100 · 500 MW = 20 MW for a frequency drop of 4 Hz and a self-
42  Power supply challenges

Figure 2.11.  Parts of Manhattan were left in the dark after hurricane Sandy in 2012.

regulating power sensitivity of the load of 1% per Hz. This means that power demand
and power supply are fully matched again at 46 Hz so that the frequency will not
decrease further.
The output of the primary reserves continues to increase after the minimum in
frequency has been reached. This additional power supply will again accelerate the
inertia. This will go faster for an inertia constant of 5 s than for one of 10 s, since
less energy is needed to bring the rotors back to nominal speed in case of lower
inertia. The deep dip in frequency observed for the lower inertia value is unaccept-

Inertia constant 5 s Inertia constant 10 s


51

50
Grid frequency (Hz)

49

48

47

46

45
0 25 50 75 100 125 150
Time (seconds)

Figure 2.12.  The effect of lowering the inertia constant on the frequency dip in case of
a calamity (further conditions as in Figure 2.11.).
2.  Balancing the electricity supply in case of calamities  43

able is most cases. If it is not possible to increase the inertia, the amount of primary
reserves has to be increased or the primary reserves have to be made faster with a
smaller initial delay.

2.2.6. The effect on frequency deviations of more powerful or faster


primary reserves.
Increasing the amount of primary control reserves and having a higher ramp rate in
the reserves help mitigate the effects of disturbances. Both measures will reduce the
undesired deep dip in frequency, and will shorten the time needed to return to 50 Hz
following the loss of a generator.
Doubling the power capacity of the primary control reserves, which results
in twice as much capacity as the lost output of a failing generator in our previous
example, reduces the dip in frequency by almost a factor of two. This is illus-
trated by the dark red line in figure 2.13. The reason that the reduction in dip is
not exactly a factor of two is that the self-regulating power decreases the load to
a lesser extent when grid frequencies come closer to 50 Hz. Another observation
is that with a higher amount of primary reserves, the ultimate quasi steady state
frequency will be closer to 50 Hz than in the case where the primary reserves
equal just the loss in power from the tripped generator. This is because even above
48.8 Hz, where the output reduction of the primary reserves kicks in because of
the chosen droop curve, more power than the power lost from the failing generator
is available now.
An additional advantage of higher primary reserves is the shorter time that the
grid frequency substantially deviates from the nominal frequency of 50 Hz. Figure
2.13 clearly illustrates that the cumulative, or integral, loss in grid frequency over
time is much lower for the red line than for the blue line. This cumulative loss is
the area between each curve and the 50 Hz line in figure 2.13. The dimension of
cumulative loss is Hz s, because it is the result of a deviation in Hz during a given
number of seconds. For the blue line, the cumulative frequency deviation equals
146 Hz s. Doubling the primary reserves brings the cumulative frequency deviation
in the example of Figure 2.13 down to only 41 Hz s. Therefore, doubling the primary
reserves decreases the cumulative frequency deviation by a factor of 3.5 in this case.
A grid operator has to ensure that, at the end of the day, the average frequency is
back to 50.000 Hz, to avoid for instance, deviations of synchronous clocks. This
means that after a frequency dip, all generators have to run at a higher frequency
than 50 Hz for a while to compensate for the dip. With a lower cumulative frequency
deviation, less effort is required to compensate for this deviation.
Should more primary reserve capacity be made available than the nominal output
of the largest generator in a system in order to avoid excessive frequency dips in case
of contingencies, a larger amount of generating capacity must be run below nominal
load. This has a negative impact on capital costs, fuel consumption, and operation
and maintenance costs.
44  Power supply challenges

Doubling the power-up ramp rate of


the primary control reserves, rendering Primary reserves 2x lost generator output
full output in 15 s instead of 30 s, results Typical point for load shedding
Primary reserves equal to lost generator output
in almost the same positive effect on the 50.5
frequency dip as doubling the capacity 50.0
of primary control reserves. This is

Grid frequency (Hz)


49.5
illustrated in figure 2.14. The only slight
49.0
difference is that the ultimate frequency
before the secondary control reserves 48.5
step in will not be slightly above 48.8 Hz. 48.0
This is the logical consequence of the 47.5
4% droop in the example: if the grid 47.0
frequency rises above 48.8 Hz, the power 0 25 50 75 100 125 150
output of the primary control reserves Time (seconds)
automatically decreases again. However,
if the droop setting of the more agile Figure 2.13.  Frequency dip in the case of primary reserves
primary control reserves is changed to with two different power capacities, otherwise the same condi-
tions as in Figure 2.11. apply.
2%, the exact same positive curve as for
doubling the primary control reserves
30 s response time,
will result. double amount
A major conclusion here is that faster 50 primary reserves
primary control reserves offer an effec-
tive option for reducing the frequency
Grid frequency (Hz)

49.5
deviation caused by a contingency. If the
Only 15 s response
relative amount of the system’s rotating 49 time, minimum
inertia decreases, such as when much primary reserves

indirectly coupled renewable electricity


48.5
sources that do not add to the inertia
are introduced, the decline in frequency
48
after a major loss in generation will be 0 10 20 30 40 50 60
faster. In that case, more primary control Time (seconds)
reserves are required to keep the system
stable, or the primary control reserves Figure 2.14.  The almost identical effect of a faster ramp-up
need to have higher ramping rates and a rate of primary control reserves compared with a doubled pri-
mary control response capacity (otherwise same conditions as
shorter initial delay.
in Figure 2.13).

2.2.7. The solution for delivering


faster primary reserves
Nowadays, agile and flexible generator sets are available that have a very fast response
to stepwise changes in the desired power output. Such smart generators can change a
certain amount of their output rapidly, with a delay of less than 1 s, the actual response
time depending to some extent on their running speed. Naturally, the allowed increase
2.  Balancing the electricity supply in case of calamities  45

in output depends on the power output set


point before the request for a change. As Best in class response for
primary frequency control Aeroderivative
an example, a machine that runs already at gas turbine
100
80% load can never accept a load increase Combustion
engine
of 40% of the nominal output. 95

(% of nominal)
Power output
Figure 2.15 shows the response of a Industrial gas turbine
90 combined cycle
number of different power generating
Steam-based
techniques to a stepwise change in 85 power plant
desired output. The data are based on
80
best-in-class machines. Less agile types
might be slower by more than a factor of 75
two. The response curve for the combus- 0 10 20 30
tion-engine-driven power plant applies Time (seconds)

for smart power generation systems


based on turbocharged engines with Figure 2.15.  Response of different generating techniques to
electromagnetic gas injection valves per a stepwise change in the power output set point .
cylinder. Faster primary reserves result
in smaller deviations from the nominal
grid frequency during calamities, even Synchronous area
where there is less inertia in the system.
One important aspect of primary Control Control
reserves is their location in the system. Block Block Control
area
If a large power plant of 1500 MW fails,
the primary reserves cannot be located
1000 km away since it would cause the
interconnecting transmission lines to
overload. This is why large synchronous
areas are split up into control blocks
and control areas. Each control area
should be able to resolve most of the
consequences of its own contingencies. Figure 2.16.  Control areas and control blocks interconnected
Neighbouring control areas in the same with high-voltage transmission lines in a synchronous area.
block are allowed to offer some support, The blue blocks represent power plants.
provided the interconnectors can carry
the load. As with most things in life, local problems should preferably be solved
locally.

2.2.8. Conclusions regarding primary reserves and inertia levels


Occurrences such as the loss of an active power plant or the loss or arrival of a major
load, will always create a disturbance in frequency. The initial change rate in frequency
is determined by the size of the unbalance between power supply and demand and by
the inertia constant of the system. Primary control reserves, i.e. power plants running
46  Power supply challenges

at part-load, adapt their output automatically when the grid frequency changes since
their desired output is determined by the grid frequency via their droop curve. Primary
reserves are preferably allocated in such a way that a control area can resolve a large part
of its own contingencies.
It is clear that opting for just a few large power plants to supply the required
electricity for an area is not ideal. In a system having a large number of generators in
a single control area, it is easier to compensate for the failure of one generator with
the other generators. The output of an individual unit is then just a small fraction of
the combined output. With a large number of generators active as primary reserves,
there is also no need to operate them at a relatively low load. In addition, the failure
of one of them has only a minor effect on the combined reserve capacity. In other
words, multiple generators in a system improve the reliability of primary reserves
and reduce the impact of a failing unit.
Fast responding primary reserves can compensate for less rotating inertia
without the need of having more primary reserves available. Without fast primary
reserves, more power plant capacity has to operate at part-load, i.e. below its rated
output. Operating at part-load increases the fuel consumption (MJ/kWh), as well as
the capital and maintenance costs per kWh (Figure 2.17). Running a generating unit
at 50% load doubles the maintenance and capital costs per kWh.

2.3. Secondary and tertiary control reserves


After activation of the primary control reserves, secondary and tertiary reserves (also
known as frequency restoration reserves (FRR), and replacement reserves (RR),
respectively) are needed. There are two reasons for this. Firstly, when primary reserves

8.0 4.5
Specific capital + maintenance
Specific fuel consumption

7.5 4.0

3.5
costs (cts/kWh)

7.0
(Mj/kWh)

6.5 3.0

6.0 2.5

5.5 2.0

5.0 1.5
100 150 200 250 300 350 400
Load (MW)

Figure 2.17.  Example of fuel consumption, and capital and maintenance costs of a
400 MW power plant according to load.
2.  Balancing the electricity supply in case of calamities  47

have been fully activated, no spare frequency control capacity is available for another
tripping power plant or a loss of major load. The risk of another power plant failing
or load disturbances taking place is always higher during a major event in the system
than when everything is running smoothly. Secondly, due to the droop characteristic
of the deployed primary reserves, a deviation from the nominal frequency of 50 Hz
will remain. Activation of the secondary control reserves will supply extra power to the
system so that the grid frequency can return to its nominal value. As a consequence,
the droop-based primary control reserves will automatically return to their original set
point and be released from their action until the next disturbance occurs.
In the Continental Europe synchronous area, the secondary control reserves
in the relevant control area automatically commence delivering output within 30
seconds following a major disturbing occurrence. This happens when the primary
control reserves are fully active. After 15 minutes, the full capacity of the secondary
control reserves has to deliver its power to the system. This approach is illustrated
in figure 2.18. Until recently, such a short deployment time required all secondary
control reserves to be spinning all the time. Large power plants can never provide
full output from standstill within 15 minutes. As a consequence, secondary reserves
based on such power plants would always be running at a level below their nominal
output. This again causes higher fuel consumption and higher capital and operational
costs.

Secondary reserves

15 min
Delay
30 sec
Ramp up

50.000 Hz
reference Δf Dead band
30 sec Power
+
– output
Ramp up

Actual grid
frequency
f + Other
– generators
ΔP ΔP

System
dynamics
System
load + loss

Figure 2.18.  A possible setup for primary control reserves and secondary control reserves in the system.
48  Power supply challenges

2.3.1. Engine-driven power plants can provide secondary reserves


from standstill
Nowadays some engine-driven power plants are able to deliver full output from standstill
within about 5 minutes (Figure 2.19). This opens up possibilities for non-spinning sec-
ondary control reserves. Reaching full speed following the start command and reaching
synchronisation with the grid takes about 30 seconds for such plants. This more than
complies with the Continental Europe synchronous area’s requirement for secondary
reserves to start 30 seconds after the event that triggered the primary reserves, and to
ramp up to full output within 15 minutes.
The ability to reach full output in 5 min-
utes is faster by a factor of 3 than what is 100

Running speed and load


demanded in the Continental Europe syn- Running speed Power output

(% of nominal value)
80
chronous area.
A quick-starting power plant has to 60
be constantly preheated to provide the 40
fast performance shown in figure 2.19.
However, such power plants generally 20
consist of multiple identical generators 0
operating in parallel. Running one of 0 60 120 180 240 300 360 420
the multiple units online for electrical Time from start command (seconds)
energy production releases sufficient
heat to keep at least 30 identical Figure 2.19.  The fast ramping up in power output from
generating units preheated. Another standstill to full load of a smart power generator.
advantage of having multiple secondary
control reserve units in parallel is the low risk of losing much allocated reserve
capacity. In case a single unit fails, only a fraction of the allocated power for sec-
ondary reserves is lost.
In some control areas in competitive markets, power plant operators can bid in
the ahead markets for offering secondary reserves. The power plants offering the
lowest price for their service will normally be selected to provide the reserves. In
other systems, the power balance in the relevant area is continuously measured with
energy flow meters, and secondary balancing is activated by a computerised control
system that sends out set-point changes to selected power plants. In these cases, fast
non-spinning secondary reserves also offer substantial advantages. Non-spinning
means using no fuel, suffering no wear, and producing no emissions.
Tertiary control reserves are activated to free the secondary reserves for the
next contingency. This can be done automatically (directly activated) or manually
(schedule activated) by the transmission system operator. Part of the tertiary control
reserves can be non-spinning. This is possible when the power plant has the rapid
response capability as depicted in figure 2.19.
2.  Balancing the electricity supply in case of calamities  49

Combined output
100

Primary reserves
80
Scheduled tertiary
Power output reserves

reserves
(% of peak output of
primary reserves)

60
Secondary
reserves

40

Direct tertiary
reserves
20

0
0 10 20 30 40 50 60
Time (minutes)

Figure 2.20  Example of the sequence in utilising primary, secondary, and tertiary control
reserves following a major occurrence.

2.4. Conclusions
This chapter has explained the delicate balance between electricity generation and
demand and the consequences of rapid disturbances in the system. Rotating inertia and
the dedicated control reserves play an important role in keeping the system balanced.
With the introduction of a substantial amount of renewable electricity sources, bal-
ancing becomes more challenging. Agile, fast-reacting generators appear to offer excel-
lent balancing duties during contingencies, even in the case of less inertia and reserve
capacity in the system.
3 Balancing power
demand and supply
when conditions
change
The demand for power changes continuously according to weather and human
behaviour. Therefore, the combined output from electricity generators in a
synchronous system has to vary all the time to match the changing demand. The
introduction of variable wind and solar energy creates additional dynamics in
matching power demand and supply. Country cases are used in this chapter to
explain the challenges of this balancing act.
52  Power supply challenges

3.1. Electricity supply differs depending on local situations


First of our country examples is Finland. It represents a country with a relatively steady
power demand and hardly any variable renewable power. As an industrialised country
in a cold climate, Finland shows a demand pattern completely different from that of a
country with hot summers or a service-based economy. Ireland serves as an example of
a service-based economy with an increasing amount of wind power capacity. Germany
is an industrialised country with large amounts of both wind and solar power. Texas is
a state with hot summers and substantial potential for solar and wind power. Finally,
California, a state with high number of solar installations – and much more to come – is
discussed briefly.

3.2. Power demand pattern in Finland exemplifying an industrialised


nation
Finland is a relatively sparsely populated Northern European country that has con-
siderable energy-based industry, such as paper mills and smelters. Winters can be
severely cold. Electricity is the most common heating system for homes, albeit nowa-
days often with energy-efficient heat pumps. Larger cities use district heating based
on the waste heat from power plants. Finland has no indigenous gas reserves and the
low population density outside the major cities excludes building an extensive gas
distribution system.
Finland has some 5.4 million inhabitants, with an average annual electricity
consumption of 15.7 MWh/capita, the highest in the world apart from Norway
with 23.2 MWh/capita (data IEA, year 2011). Ranking third and fourth in annual
electricity consumption per capita are Canada with 15.7 MWh and the USA with

15 000
Hourly data power demand Finland

12 500
Midsummer
celebration
10 000
(MW)

7 500

5 000

2 500

0
0 5 10 15 20 25 30 35 40 45 50
Week numbers year 2012

Figure 3.1.  Hourly data of electric power demand in Finland for the year 2012 (hourly data from ENTSO-E).
3.  Balancing power demand and supply when conditions change  53

13.4 MWh. By comparison, the world average in annual electricity consumption per
capita is only 2.9 MWh.
Figure 3.1 shows how electricity demand in Finland varied during the year 2012.
The curve is blurred because of daily variations in electricity demand. An anomaly in
the curve is the deep dip in power demand in the last week of June. This is caused by
the famous Midsummer Day celebrations. Only pubs, restaurants, and hospitals are
functioning then. At the end of the year, between Christmas and the New Year, most
industrial and commercial activities are also down. At that time of the year, Finns are
typically all eating ham and drinking glögi, a spicy glühwein (mulled wine). Figure 3.1
reveals that Finland had a minimum electricity demand of about 6.5 GW in 2012.
Finland has a policy aimed at reducing greenhouse gas emissions and decreasing
the country’s dependence on fossil fuels. The plan is to reduce greenhouse emissions
by 80% of the 1990 level by the year 2050. Biomass, wind power and geothermal
heat should be the major sources of renewable energy. Because of the high baseload
level, the Finnish authorities have decided to support building additional nuclear
power plants. Nuclear power will replace some old coal-fired power plants and pro-
vide additional capacity to satisfy the growing demand for electricity. Nuclear power
plants have low fuel costs and very low carbon dioxide emissions. The high invest-
ment costs, however, require running them for a large part of the time at full output.
The Olkiluoto 3 power plant, which has been under construction since 2005, will
have a power capacity of 1600 MW and the planned Olkiluoto 4 plant a capacity of
between 1000 and 1800 MW. The discussions about primary and secondary reserves
in Chapter 2 made clear that installing such large power plants requires a substantial
amount of primary, secondary and tertiary control reserves for maintaining stability
in case of a trip.
The hourly data of electric power demand in Finland, as shown in Figure 3.1 is
plotted in a distribution curve in figure 3.2. Such a distribution curve is based on a
rearranged data series, starting with the highest value and ending with the lowest. It
gives a good indication as to what fraction of the total time a certain power demand
is required. Apparently, the power demand was higher than 6.5 GW for 99% of the
time. Demand exceeded 10 GW for only 40% of the time. Peaks in demand higher
than 12 GW occurred only 10% of the time. Therefore, power plants responsible for
generating such peaks in demand will have a very low utilisation factor. A utilisation
factor of 1 (= 100%) means that the power plant is producing its nominal output
continuously throughout the year. In practice, however, power plants require main-
tenance and they sometimes trip. In addition, they have to provide reserve capacity.
A utilisation factor of 0.9 for a baseload power plant is, therefore, already quite good.
The short-duration absolute peak in demand from 14.0 GW to 14.5 GW shown in
Figure 3.2 occurs only in the case of extremely cold weather. The capital costs per
kWh of power plants providing such a peak are very high. It would be better to offer
Finns a good price for reducing demand during very cold days. Maximising the use
of available wood burners and avoiding heating electric saunas would reduce elec-
tricity consumption during spells of very cold weather.
54  Power supply challenges

15 000
Distribution curve power demand

P > 12 GW, 10% of the time


12 500
P > 10 GW, 40% of the time
10 000
Finland (MW)

7 500
Base load 6.5 GW 99% of the time
5 000

2 500

0
0 10 20 30 40 50 60 70 80 90 100
% of the time, year 2012

Figure 3.2.  Power demand in Finland distributed over the 8784 hours of the year 2012,
with maximum demand on the left-hand side and minimum demand on the right.

The dynamics in electricity demand in Finland are such that conventional


power plants, in combination with imports and exports, can easily handle them. It
is not expected that Finland will install many solar panels.
However, there are plans to install 3 GW of wind turbines.
Using biomass as a renewable fuel is a good option, since
there are an estimated 5 000 to 10 000 trees per capita
in the country. Biomass based electricity production is
controllable and can, therefore, be seen as a dispatchable
supply source for the electricity grid.
Figure 3.3 combines the daily consumption of electrical
energy (GWh) and the daily maximum and minimum
temperatures at Helsinki–Vantaa airport in 2012. Low
temperatures substantially increase the use of electricity.
On February 3, the coldest day of the year, the minimum
temperature Tmin at Helsinki airport was –31 °C and
electricity consumption reached the absolute peak of 330
GWh. Demand varied that day between 12.5 and 14.5
GW. Although there is a relationship between electricity
consumption and the number of daylight hours, space
heating is primarily responsible for the high peak demand
for electricity in Finland. A cold spell in the beginning of
December 2012 also created a peak in electricity demand.
Figure 3.4 relates the use of electricity to the average
daily temperature determined from (Tmax+Tmin)/2 as given Typical Finnish landscape.
3.  Balancing power demand and supply when conditions change  55

Year 2012
February 3
350 35
Tmax (deg C)
Finland daily electricity use (GWh)

300 25

250 15

200 5
Electricity use
150 –5

100 –15
Tmin (deg C)
50 –25

0 –35
1 29 57 85 113 141 169 197 225 253 281 309 337 365

Figure 3.3.  Daily electricity use and daily temperatures in Finland in 2012.

in Figure 3.3. The relationship proves to be quite significant, witness the high cor-
relation coefficient of 0.85. It shows that a variation of between 175 and 325 GWh
in the daily electricity consumption in Finland relates to the weather. The peculiar
circular deviation in Figure 3.4 at the minimum electricity consumption of close to
150 GWh per day is again caused by the Midsummer celebrations.
Figure 3.5 illustrates the electricity demand pattern based on the hourly data
from the week commencing October 8, 2012, when the weather was still not
extremely cold and the holiday season
is over. The baseload of about 8 GW
determines the bulk of demand during
Finland daily electricity use (GWh)

350
that whole week, confirming that con-
300
tinuous industrial activities dominate
electricity use. Comparing the pattern 250
Black trend line
for week days with that of weekends 200 y = –3.6804x + 250.61
R = 0.8529 2

reveals that only part of the interme- 150


diate load is related to work activities. In
100
particular, the peaks at the end of every
50
day are caused by domestic activities.
Typical Friday evening and Saturday 0
–30 –20 –10 0 10 20 30
evening peaks are caused by the almost
Average temperature Helsinki-Vantaa airport (°C)
compulsory weekly family sauna. There
are around 3 million saunas in Finland
and the majority of them are electrically Figure 3.4.  Correlation of the average daily temperature at
heated. Helsinki airport with daily electricity use.
56  Power supply challenges

Figure 3.6 gives the power demand


pattern for Monday October 8, 2012. It
reveals that Finns start their activities
early in the morning and go to bed early
in the evening. Although an increase in
demand from 8 GW to 10 GW between 7
am and 9 am looks small in the diagram,
it is still a substantial amount of addi-
tional power. The maximum ramp-up rate
in power demand is about 1 GW/hour.
Starting already from noon, the power
demand slowly decreases on weekdays.
The intermediate load level between 7 am
and 8 pm covers a continuous time span Modern Finnish sauna.
of 15 hours. The demand pattern in Figure
3.6 does not show substantial peaks. Week 41, 2012
In conclusion, Finland is a country 12 000

with a relatively high season-dependent


Power demand Finland

10 000
baseload. The daily variations in elec-
8 000
tricity demand are about 10% of the
(MW)

6 000
average demand. Daily cycling in demand
is quite predictable and the ramping up 4 000
and ramping down can be easily covered 2 000
by early preparation of the power plants.
0
Finland has about 3.3 GW of hydropower, Mo Tu We Th Fr Sa Su
which helps in shaving the peaks and in
ramping electricity production up and
Figure 3.5.  Electric power demand in Finland during the
down.
week of October 8–14, 2012 (hourly data from ENTSO-E).

3.3. Power demand in the


Republic of Ireland, exemplifying Monday, October 8, 2012

a system with much wind-based 12 000


Power demand Finland

power 10 000
8 000
(MW)

Ireland has been chosen as an example 6 000


because it has a notable amount of wind 4 000
power in its electricity supply system. Like 2 000
Finland, Ireland is a country with a rela- 0
tively small number of inhabitants, but it 0 2 4 6 8 10 12 14 16 18 20 22 24
is representative of those regions with a Hour of the day
moderate climate and a limited amount of
industry. An interesting aspect is that Ire- Figure 3.6.  Electric power demand in Finland on Monday,
land aims to install substantially more wind October 8, 2012.
3.  Balancing power demand and supply when conditions change  57

power. A target of between 5.5 and 6.3 GWh is mentioned in the Irish Power plant capacity Finland in
Gate 3 project for covering ultimately 40% of the electricity demand 2010: 17 GW
from renewables by the year 2040. The country has some 4.5 million
inhabitants and an electricity consumption of 4.5 MWh per capita. Renewables
and peat
Figure 3.8 shows that the dynamics in electricity demand in Ireland 15%
are proportionally much higher than those in Finland. This is typical Fossil
Nuclear
for a service-based economy. The baseload is only about 2 GW, com- 16% fuels
49%
pared with at least 7 GW in Finland. The daily variations in demand are
also close to 2 GW, about the same as in Finland. The peaks and val- Hydro
leys in electricity demand appear to follow a well defined pattern. The 20%

only anomaly, i.e. the dip in demand at the end of the year, is apparently
caused by Christmas and the related holiday week.
The distribution curve of power demand in Ireland (Figure 3.9) Figure 3.7.  Finland has a substantial
shows that the peak between 4 GW and 4.6 GW occurs for less than amount of hydropower, which creates
2.5% of the time and is, therefore, very uneconomic from a view- flexibility.
point of capital costs for the installed generating capacity. A base
demand of almost 2 GW means that having just a few large power plants to cover
this baseload creates a high risk of blackouts in case of a calamity. Figure 3.10 gives
the power plant portfolio in Ireland for the year 2010. The total generating capacity
slightly exceeds 6.2 GW, which is enough to cover the maximum demand. Seven
gas-fired power plants dominate the portfolio. The largest power plant has a capacity
of 463 MW. If that one trips during the night and its full output is lost, about a
quarter of the 2 GW online generating capacity disappears. It is close to impossible
to correct such a contingency with primary control reserves. The addition of flexible
power plants based on multiple units in parallel with an intrinsically quick response
to demand changes might offer a solution here.
Power demand in Ireland is clearly seasonal, with the dark seasons requiring more
power than when there is more light. Home heating in Ireland is traditionally based

5 000
Power demand Ireland (MW)

4 000

3 000

2 000

1 000

0
0 5 10 15 20 25 30 35 40 45 50
Weeks year 2012

Figure 3.8.  Power demand in the Republic of Ireland in 2012 (data from Eirgrid)
58  Power supply challenges

on oil, although natural gas is replacing


oil in some areas. The Gulf Stream in the 5 000

Power demand Ireland (MW)


Atlantic Ocean has a moderating effect on P > 4 GW, 2.5% of the time
the climate. The average daily temperature 4 000
varies between 5 and 16 °C throughout 3 000
the year, a much smaller range than the
–20 to +20 °C in Finland. 2 000 Base demand ≈ 2GW
Daily power demand variations in
1 000
Ireland have the same characteristics as
in other countries having service econo- 0
0 10 20 30 40 50 60 70 80 90 100
mies and moderate climates, such as Bel-
% of the time
gium and The Netherlands. Demand is
always lower on Saturdays and Sundays
than it is on weekdays. Figure 3.9.  Distribution curve of the power demand in the
On a weekday, power demand starts Republic of Ireland during 2012.
to rise from around 6 am and reaches
a steady level by about 10 pm. The black curve in Figure 3.13 shows the demand
pattern on Monday February 27 only. This was a day with high electricity demand.
In Figure 3.11 the red curve shows that the maximum rise in power demand per
half hour was about 260 MW between 7 am and 8 am. A slightly smaller rise rate
in power demand occurred around 7 pm, when people arrive home, switch on
the lights and prepare their evening meal. The fastest ramping down took place
between 10 pm and 11 pm, at a rate of 200 MW/30 min.
With Atlantic Ocean winds, Ireland is suitable for wind turbines, especially at its
west coast. The amount of installed wind turbine capacity in 2012 averaged approxi-
mately 1650 MW.
The combined power output from Irish wind turbines shows high levels of vari-
ability, notwithstanding the fact that the capacity is spread over some 125 wind

Irish fuel-based electricity generation capacity (MW)


Nominal power capacity (MW)

500
450
400
350
300
250
200
150
100
50
0
Heavy Fuel Oil
Heavy Fuel Oil
Heavy Fuel Oil

Heavy Fuel Oil


Heavy Fuel Oil
Heavy Fuel Oil
Heavy Fuel Oil
Distillate Oil

Distillate Oil
Distillate Oil

Distillate Oil
Distillate Oil

Gas Gas Gas Gas Gas Gas Peat Gas Gas Peat Gas Coal Coal Coal Gas Gas Gas Gas Gas Gas Peat Gas

Reeks1 258 90 90 90 432 403 118 111 54 54 108 342 400 91 85 283 283 283 163 104 463 52 52 81 81 54 54 241 241 52 52 384 137 445

Figure 3.10.  The fuel-based power plant portfolio in Ireland (year 2010).
3.  Balancing power demand and supply when conditions change  59

parks. Figure 3.14 reveals that the power


output from the combined wind turbines

Temperatures and daylight hours


20 3.0

Monthly electricity use (TWh)


varied between 1 500 MW and zero, with
a tendency towards less energy produc- 15 2.5
tion during the middle part of the year.
Yet, in 2012 output equal to the installed 10 2.0
power of 1650 MW was never reached.
Although the variability in wind power 5 1.5
output is substantial, the higher output
during the darker seasons when more 0 1.0
power is needed than in the summer is 0 60 120 180 240 300 360
a positive thing. Nevertheless, the com- January 1 – December 31, 2012
bined output from the wind turbines is
Min temp (°C) Max temp (°C) Electricity use Daylight hours
also often close to zero.
Figure 3.16 gives the cumulative elec-
tricity production of the Irish wind parks Figure 3.11.  Monthly electricity use in Ireland compared with
for 2012. The inclination of the curve is the number of daylight hours and the average minimum and
maximum temperatures at Dublin airport.
lower in the middle of the year than early
and late in the year. This confirms the
February 27 – March 4, 2012
lower electrical energy production from
Ireland power demand (MW)

5 000
wind power in the summer time. The total
4 000
amount of electricity produced by the
combined windmills during the leap year 3 000
2012 was 4.1 PWh. A leap year has 8 784 2 000
hours and the average installed wind tur-
1 000
bine capacity was 1650 MW. This renders
a capacity factor of 4100000 MWh/(1 650 0
MW · 8784 h) = 0.28, or 28%. This is quite
good for wind energy. In comparison, the Figure 3.12.  The electricity demand pattern in Ireland from
capacity factor of the German wind parks Monday February 27 to Sunday March 4, 2012.
was just 17.5% in 2012.
Total electricity demand in Ireland
Ireland power demand (MW)

was 25.6 PWh = 25600 GWh in 2012, so 4 000 300


dP/dt (MW/30 min)

the wind turbines covered 4.1/25.6 · 100% 3 500 150


= 16% of demand. This, however, does not
3 000 0
mean that less power capacity from the
other generators was required, since there 2 500 Ramping up
–150
260 MW/30 min
are times when the combined windmills 2 000 –300
0 2 4 6 8 10 12 14 16 18 20 22 24
have hardly any output at all. Figure 3.16
Hour of February 27, 2012
shows that the wind-based power output
was lower than 165 MW. i.e. 10% of the Figure 3.13.  Electricity demand pattern (black) in Ireland on
installed capacity, during 23% of the time, February 27, 2012, with the half-hourly change rate dP/dt in
demand (red).
or for 84 days, in 2012. This means that
60  Power supply challenges

Year 2012, 1 650 MW wind turbine capacity in Ireland


1 500
Wind turbine power output

1 250
1 000
(MW)

750
500
250
0
0 5 10 15 20 25 30 35 40 45 50
Weeks of 2012

Figure 3.14.  The combined output from wind turbines in Ireland during 2012 (data from
Eirgrid).

fuel-based capacity was largely responsible for covering demand during that fraction
of the time. The output from wind turbines exceeded 1 000 MW for only 10% of the
time, or 36 days.
The installed fuel-based electricity generating capacity in Ireland is 6.2 GW,
which would render a capacity factor of 25 600/(6.2 · 8784) = 47% if the power
plants had to produce the full 25.6 PWh in demand. Because wind produced
4.1 GWh, the power plants delivered only 21.5 GWh. Their actual capacity factor
was therefore 39%. The blue curve in Figure 3.17 shows that due to wind power, the
baseload for the power plants is some 1.5 GW, considerably lower than the 2 GW

Figure 3.15.  Ireland plans to double the amount of wind energy by 2020.
3.  Balancing power demand and supply when conditions change  61

base demand. Yet, the peak output from


Wind power output distributed curve for 2012
power plants still equals peak demand: 1 650 MW installed, capacity factor 28%
4.6 GW. Since the same amount of power 1 500
Power only 10% of

Wind turbine power output


plant capacity is needed as when there is the time > 1 000 MW
1 250
no wind power, the consequence is that
1 000
the utilisation factor of the power plants

(MW)
decreases by a factor of 47/39 = 1.2. This 750
Power 23% of
means that the capital costs for the power the time < 165 MW
500
plants per kWh produced are 20% higher
250
than in the case of no wind power.
The fact that wind turbines produced 0
0 10 20 30 40 50 60 70 80 90 100
16% of the electrical energy needs in Ire- % of the time
land in 2012 does not mean that the fuel
consumption for covering the electricity Figure 3.16.  Distribution curve of the output from the com-
needs was 16% lower than if there were bined wind turbines in Ireland during the 8 784 hours of 2012.
no wind power. Because of the volatile
and unpredictable character of wind Ireland 2012: effect of wind power output on
output, fuel-based power plants will, on power plant output distribution
average, run at a relatively lower frac- 5 000

tion of their nominal capacity than they 4 000


Power (MW)

Power demand
would if there were no wind turbines 3 000
in the system. The consequence of this Power plant output
2 000
is higher fuel consumption and higher 1 000
maintenance and operational costs per
0
kWh. The 2012 load-following pattern of 0 10 20 30 40 50 60 70 80 90 100
the power plants is shown in figure 3.18. % of time
This is a completely different picture than
the power demand curve in Figure 3.8, Figure 3.17.  Comparison of the distribution curves of power
since the variability is much higher. Ire- demand and demand minus the contribution from wind turbines.
land’s typical base demand of 2 GW has
translated into a base output of only 1.5 GW from the power plants. Consequently,
many more starts and stops and higher ramping up and ramping down rates occur
for the power plants that provide intermediate and peak load than without wind
power in the system. This variability increases the wear rate of the power plants.
To illustrate the level of increased volatility imposed on power plants due to
wind energy, the change in power demand per half hour and the change in power
supply from the fuel-based generators have been plotted in figure 3.19. At first sight,
the difference between the two diagrams is not so large. Both diagrams are still dom-
inated by the weekly demand patterns. However, one can notice that the maximum
in change rate for the power plants is 450 MW/30 min, while power demand has a
maximum change rate of 350 MW/30 min. Ramping up the electricity supply is in
both cases higher than ramping down. However, since less fuel-based power plant
capacity is online because of the output from wind power, the proportion of volatility
62  Power supply challenges

Power demand minus the output from wind turbines


5 000
Power supply power plants

4 000

3 000
(MW)

2 000

1 000

0
0 5 10 15 20 25 30 35 40 45 50
Weeks of 2012

Figure 3.18.  Half-hourly values of the power demand minus the wind power output
during 2012, representing the contribution from non-wind power plants to power demand.

per the level of power plant output is much higher. Nevertheless, deriving just 16%
of the electrical energy from wind power having a capacity factor of 28% does not
create serious balancing problems.
When the output of power plants has to increase 300 MW within a time span of
half an hour, it makes considerable difference when only 1000 MW is online than
when 2000 MW is online. Reserve capacity for contingencies and forecasting errors
will help in ramping up output, but where only little power plant capacity is online
in cases where fast ramping up is needed, this hardly helps. Without wind turbines,
the change in power demand from power plants would never have had ramping up

1 vertical unit = 200 MW/30 min


Demand
Half-hourly changes
(MW/30 min)

Power plants

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
January 2012

Figure 3.19.  Comparison of the half-hourly change rates in power demand and supply
from the fuel-based power plants (situation year 2012).
3.  Balancing power demand and supply when conditions change  63

Double wind power compared with 2012


5 000
4 000
Power supply from
power plants

3 000
(MW)

2 000
1 000
0
–1 000
0 5 10 15 20 25 30 35 40 45 50

Weeks of 2012

Figure 3.20.  Half-hourly values of the calculated power supply from Irish power plants
with double the wind power capacity of 2012.

values higher than 14.5% of the online power-plant capacity per 30 minutes in 2012.
Ramping down would never have been faster than 9.6% of the online power-plant
capacity per 30 minutes. For the wind power capacity in 2012, the power plant
ramping up showed a maximum of 23% of the online power-plant capacity and
ramping down 16.3%. However, for a doubling of wind power, the power plants will
often be further pushed from the grid. In that case, the relative changes in power
requested from them will be high, exceeding 500% of the online capacity. Conse-
quently, a limited amount of wind power in a system is factors easier to accommo-
date than a large fraction.
In the case of a doubling of wind turbine capacity in Ireland compared with that
of 2012, the half-hourly changes in power supply for the fuel-based generators would
occasionally exceed 550 MW. Figure 3.20
illustrates that the backup pattern of the Effect of doubling wind power output on power
fuel-based power plants would be totally plant output distribution
5 000
different from the power demand pattern
shown in figure 3.8. The baseload part 4 000
Power
Power (MW)

for the power plants almost disappears. 3 000


demand
This disappearance is a consequence of Power plant
output
deriving more than 25% of the electrical 2 000
Demand-
energy from wind. The power output doubled wind
1 000
from double the amount of wind turbines
0
would exceed demand at least 15 times 0 10 20 30 40 50 60 70 80 90 100
per year. % of the time
Peak demand from fuel-based power
capacity does not, in practice, decrease Figure 3.21.  Distribution curves for 2012 showing the impact
when the installed capacity of wind tur- of wind power output on the annual distribution of fuel-based
bines increases from 1.65 GW to 3.3 GW, power plant output.
64  Power supply challenges

Table 3.1.  Effect of wind power on fuel-based power plants in Ireland,


based on year 2012 data.

Power Power Power plant Power plant Fraction Utilisation


plant max plant min average demand factor
Situation crest factor covered by power
output output output
wind plants

MW MW MW max/ % %
average
No wind
power 4588 1624 2917 1.57 0 47
2012 wind
power 4446 913 2450 1.81 16 39

Doubled
2012 wind 4384 –207 1980 2.21 32 32
power

witness figure 3.21. In particular, the relative volatility in output of the fuel-based
power plants will increase and many more starts and stops will be experienced. In
addition, the capacity factor of the existing power plants will decrease further to
32%.
Table 3.1 summarises the effects of wind power on the fuel-based power plants
in Ireland. The three scenarios of no wind turbines, the 2012 installed power of wind
turbines, and a doubling of the 2012 installed power of wind turbines are shown. The
power plant crest factor is the peak load of the power plants divided by their average
load. The crest factor is an indication of the volatility in output that the power plants
experience. The minimum power plant output can never be less than zero; the ‘– 207
MW’ in the third column of table 3.1 only indicates that the wind turbines will occa-
sionally produce more electricity than is requested by demand.
Before the introduction of large levels of wind and solar-based electricity supply,
the conventional power plants in Ireland were generally able to follow demand
because of its high predictability.
Sophisticated models could predict the
load quite well based on historic data 42.23
and weather forecasting. Local winds 40
and sunshine are far harder to predict,
Distribution (%)

and even forecasting the average wind 30


speed for a country such as Ireland does 20.67 20.29
20
not always give accurate results. Figure
3.22 gives the deviation between the 6.94
10 5.12
forecast for the total wind power output 0.13 0.89
2.43
0.80 0.52
in Ireland and the actual output, based 0
on data from Eirgrid for the year 2012. –550 –450 –350 –250 –150 –50 50 150 250 350 450 550

The deviations have been arranged in Wind power forecast minus actual wind power (MW)
11 classes of 100 MW each (Figure
3.22). In 42% of the half-hourly data, Figure 3.22.  Distribution of differences between day-ahead
the deviations amounted to between +50 forecasted wind power and actual wind-power output in Ireland
MW and –50 MW. For almost 7% of the for 2012 (classes of 100 MW).
3.  Balancing power demand and supply when conditions change  65

time, the forecast is between 150 and 250


MW too high. There are even a few cases 1 600

Actual wind power output


where the forecast exceeds the actual 1 400 y=0.8676x + 50.291
wind output by more than 450 MW. r 2 = 0.8975
1 200
This means that additional generating 1 000

(MW)
capacity has to be reserved to fill the gap. 800
This reserve capacity has to be fast, pro- 600

viding output changes of between + and 400

–100 MW in 30 minutes, with sometimes 200


0
extremes of –550 MW and +350 MW.
0 200 400 600 800 1 000 1 200 1 400 1 600
Such excursions in unexpected output
Forecast wind power output (MW)
require special generating machinery.
Figure 3.23 illustrates the difference
between the forecasted wind power Figure 3.23.  Actual wind power output versus day-ahead
output and the actual output from the forecasted output (data from Eirgrid).
combined wind turbines in a different
way. The data points for the actual output of the combined wind turbines are plotted
against the forecasted data points. The red trend line shows that the forecast is on
average 10% higher than the actual output. Occasional large deviations of around
500 MW between the forecasted output and the real output are clearly visible in this
diagram.
In a system where more than 25% of the electricity demand is covered by wind
power, the portfolio of power plants has to be flexible, able to provide fast ramping
up and down, and be capable of enduring frequent starts and stops. The challenge
for balancing increases when the capacity factor of wind parks decreases. The reason
for this is that the ratio between peak output and average output is higher for a lower
capacity factor. In order to comply with the rules for contingency reserves, the unit
size of the power plants should be limited. Although the median value of the supply
from power plants for a doubling of the wind capacity in the Irish system in 2012
is still 2 GW, power plants often have to deliver less than 1 GW. This 1 GW should
never be provided by less than 10 generating units, since otherwise a failure of one of
the units would require an excessive load step for the remaining units.
Electricity from an onshore wind park costs about 10 €cnt/kWh, given a 5%
interest on the investment, a technical life of 20 years, and a capacity factor of
about 30%. If, in the Irish case, the 2012 wind park capacity is doubled to pro-
vide 32% of the electricity demand, the kWh costs of the non-wind power plant
portfolio might increase by about 3 €cnt due to a lower utilisation factor, higher
fuel consumption per kWh, and a higher rate of wear. In order to show the actual
integrated costs of wind energy, these 3 €cnt/kWh should be allocated to the wind
energy costs via the fraction of the power plant electricity production relative to
the wind turbine production (100–32)/32 = 2.1. The real costs of the Irish wind
electricity would, therefore, be about 10 + 2.1 · 3 = 16.3 €cnt/kWh should the
2012 installed capacity be doubled. This does not take into account the costs of
66  Power supply challenges

Wind output major wind energy countries Europe


40 000
35 000 Spain
Wind power output (MW)

30 000 Germany rest


25 000 DK east
20 000 DK west

15 000 Germany NW

10 000 Germany NE
Ireland
5 000
0
31 days of January 2012

Figure 3.24.  Interconnecting the European wind turbines does not help to smooth their
electricity production, because atmospheric conditions are often continent-wide.

the extra reserve power required because of


deviations between the predicted and actual
wind-power output. However, if the wind tur-
bines are produced and maintained by the Irish
themselves, they will also be job creators. More-
over, an interest rate of 5% is nowadays quite
attractive for private investors. The benefit or the
burden on the economy of a large-scale applica-
tion of wind turbines therefore depends on the
boundary conditions in the country.
A basic negative aspect of wind parks is
the variability in output. The widespread idea Figure 3.25.  An example of a weather system domi-
that interconnecting European wind parks with nating a continent.
high-voltage lines, the so-called super grids or
electricity highways, will smooth the peaks of
wind-power can be shown to be incorrect. Figure
3.24 gives the combined output from all the wind
turbines in Germany, Denmark, Ireland and Spain
in 30 minute intervals in January 2012. The com-
bined output is very peaky. The reason for this is
that weather systems in Europe generally cover a
large area of the continent.
In summary, a transition to an electricity
system where wind power covers more than 25%
of the demand means that the large conventional Figure 3.26.  Adjustment of wind turbine output by
power plants cannot alone ensure the delicate bal- pitching of the blades and yawing of the nacelle.
3.  Balancing power demand and supply when conditions change  67

ancing of demand and production. Fast hydropower and pumped hydro might help
but such facilities are often not present in sufficient levels. Shaving peaks in output
from wind turbines with electrical heaters in hot water boilers might help to some
extent. Another method is to curtail high outputs from wind turbines by yawing and
pitching of the propellers. Yawing means rotating the nacelle away from the position
where it catches most of the wind energy, while pitching means changing the angle
of the blades for the same purpose. Many countries, however, allow renewable energy
sources unrestricted supply into the grid. Ultimately, flexible, agile power stations are
needed for load balancing, for contingency reserves, and for providing reserve power
should the actual wind power deviate from the forecasted value.

3.4. The 50Hertz transmission system operator region in Germany,


a region with much solar-based power
Electricity produced by photovoltaic panels irradiated by the sun shows variability,
but the nature of the variations is very different from that of wind-based energy. The
output from solar PV panels is always zero during the night, but an overcast sky will
also heavily reduce their output. During the day, the output of a single PV panel can
drop rapidly from 100% to 10% should a cloud suddenly block the solar radiation.
This means that clusters of solar PV panels can show much more drastic changes in
output than wind farms. PV applications on smaller islands and autonomic municipal
electricity supply systems in particular can suffer from this. However, combining the
output from solar PV panels covering a widespread area will smooth the variations
caused by clouds.
Figure 3.27 shows the power output from the combined solar PV panels in the
area operated by the 50Hertz Transmission System Operator in Germany in the week
commencing May 20 (week 21), 2013. In
early 2013, the installed PV capacity was
already 7.2 GW, following an increase of Solar PV output, 50 Hertz Transmission,
week 21, 2012
3 GW during 2012. The installed thermal 5 000
power plant capacity was 17.8 GW and week 21, 2013 forecast
week 21, 2013 actual
4 000
Power output (MW)

the wind-turbine capacity 12.4 GW. The


pumped-hydro capacity was 2.8 GW. The 3 000
50Hertz Transmission System Operator
has a relatively large share of PV capacity 2 000
installed in its region, and the 15-minute
1 000
interval output data are available via the
internet. 0
Clouds are the cause of the daily dif- Mo Tu We Th Fr Sa Su

ferences in Figure 3.27. The output can


be different by a factor of four. Clouds are Figure 3.27.  Power output from the combined solar PV
also the reason why the output curves are panels in the 50Hertz Transmission GmbH area in Germany
not always smooth. The forecast for the during the week commencing May 20, 2013.
68  Power supply challenges

Solar PV output, 50 Hertz TSO, Germany


Daily electricity production (GWh)

40
35
30
25
20
15
10
5
0
0 5 10 15 20 25 30 35 40 45 50
Weeks of 2012

Figure 3.28.  The daily electricity production from the solar PV panels in the 50Hertz
transmission system operator area in Germany in 2012.

combined output of the PV panels does not deviate excessively from the actual value.
Although the installed PV capacity (peak capacity) in the region must have been close
to 8 GW in week 21, 2013, that value was never reached. The major reason for this is
the non-optimum position of many panels and the presence of clouds in the area.
Since the latitude of the 50Hertz transmission system operator is between 50 and
54 north, solar irradiation from October to February is quite low. Figure 3.28 gives
the daily electrical energy production from the solar panels in that area expressed
in GWh for the 366 days of 2012. The diagram does not represent the actual solar
irradiation since the installed PV panel capacity almost doubled during the year.
Nevertheless, the output per day can
vary by a factor of 500. In the darker 50 Hertz Transmission region Germany,
seasons, the contribution from the solar year 2012
panels is negligible. This means that the 5 000
May-25
capacity of the other power plants cannot
Solar PV output (MW)

4 000
be reduced as long as no economic long-
term energy storage facility for solar- 3 000
power is available.
2 000
Figure 3.29 compares three very Jun-21
different days of solar PV electricity 1 000
production. June 21 could have shown Jan-02
the highest output from the solar panels, 0
but apparently the weather situation was 0 4 8 12 16 20 24
not favourable for PV. The fluctuations in Time of the day
PV output in the time span around noon
on June 21 and May 25 are also caused Figure 3.29.  Three days in 2012 having substantial
by clouds. differences in PV panel output.
3.  Balancing power demand and supply when conditions change  69

Solar PV panels in areas in latitudes


above 40° have a very low output in the Statistical distribution solar PV output, 50 Hertz TSO,
Germany in 2012

Solar PV power output (MW)


darker seasons. Solar panels can be fuel 5 000 Average installed capacity ≈ 6.2 GW
savers in the lighter seasons, but the Maximum output: 4.5 GW
4 000 Energy produced: 5.13 TWh
capacity of the other generators has to be Capacity factor: 9.5%
3 000
large enough to cover the peak electricity Output only 10.8% of the time > 1/3 of
the installed capacity
demand in the darker season. During 2 000
the course of a year, the panels have no 1000
output at all half the time, as shown in the
0
distribution curve of Figure 3.30. This is 0 10 20 30 40 50 60 70 80 90 100
caused by the absence of sunshine during % of the time
the night. As a consequence, the capacity
factor of solar panels in the 50Hertz TSO Figure 3.30.  The distribution of solar PV output in the
area does not even reach 10%. The reason 50Hertz TSO area in 2012.
in part is the positioning of the panels.
Panels installed on roofs facing west or Day-to-day energy output wind 50 Hertz TSO
east, as well as panels with an inferior 250
Daily electric energy

y = 0.0812x2 – 4.7933x + 101.86


inclination, will never reach their nominal 200 R2 = 0.1786
output. Optimum output is obtained if
150
(GWh)

the PV panels are mounted on a tracking


system that automatically selects the best 100

position of the panels with respect to the 50


sun. However, municipal planning rules 0
0 5 10 15 20 25 30 35 40 45 50
often prohibit the installation of such sys- Weeks of 2012
tems since they are ground based, while
no planning restrictions exist for roof Figure 3.31.  Daily electric energy output (GWh) from wind
mounted PV panels. turbines in the 50Hertz TSO area during 2012.
Germany also has an enormous
number of wind turbines installed. The
country’s combined installed wind turbine 50 Hertz transmission area, Germany in 2012
capacity amounted to some 37 GW by 40
Daily energy output wind

35
the end of 2012. In the 50Hertz TSO
30
area, the wind power capacity increased
turbines (GWh)

25
from 11.8 GW to 12.4 GW in 2012. It 20
y = –0.0804x + 18.069
would be ideal if the lack of sunshine in 15 R2 = 0.1285
the darker seasons could be compensated 10
for by increased output from the wind 5
turbines. However, although the peaks in 0
0 50 100 150 200 250
wind turbine output are actually higher in % of the time
the winter, there are still many days with
very little wind. The correlation coefficient Figure 3.32.  Daily energy output from the solar PV panels
between the black trend line and the actual plotted against that of the wind turbines (data points of 366
days, year 2012, 50Hertz TSO region, Germany).
data of wind energy in Figure 3.31 has the
70  Power supply challenges

50 Hertz transmission, Germany, May 25, 2012

10 000
Demand

8 000
Others
Power (MW)

6 000

4 000
Solar
Wind
2 000

0
0 4 8 12 16 20 24
Time (hours)

Figure 3.33.  Power demand, output from solar panels and wind turbines, and the resulting
balancing by the other generators.

low value of only 0.18. This means that the volatility in wind-based electricity produc-
tion is so high that the trend line gives no proper indication of the actual instantaneous
production. Even when electricity from wind and solar is combined, there are many
occasions when other generators have to cover the full electricity demand.
During times of really high wind turbine output, there is generally little sunshine.
Stormy weather and heavy clouds usually come together. Figure 3.32 shows that if
the daily electricity production from wind exceeds 150 GWh, the PV panels generally
deliver relatively little energy. During days with much sunshine, with a combined PV
panel output higher than 25 GWh, the wind-based electricity production does not
exceed 80 GWh. The blue trend line in Figure 3.32 has a very low correlation with
the actual data, meaning that no real relationship between solar PV output and wind-
turbine output is present.
Two examples will now show the impact of solar panels and wind turbines on
daily load following in the 50Hertz TSO area. May 25, 2012, was a day with very
much sunshine and moderate winds. The blue curve in Figure 3.33 shows the
electricity demand for the whole day in 15 minute intervals. Minimum demand
equalled 7 GW. Maximum demand was 10.5 GW. This is a typical pattern for an
industrialised country. The maximum rise rate in demand was about 1 GW per hour
between 5.30 am and 7 am. May 25 was a Friday, and apparently many industrial
and commercial activities closed early on that day, hence the decline in demand in
the afternoon. Satisfying such a demand pattern with conventional nuclear and coal-
fired power plants is not difficult because of the high predictability of the demand
pattern and the relatively high baseload.
3.  Balancing power demand and supply when conditions change  71

50 Hertz transmission, Germany, May 12, 2012

10 000

Demand
8 000
Power (MW)

6 000

4 000
Solar Others
Wind
2 000

0
0 4 8 12 16 20 24
Time (hours)

Figure 3.34.  The impact of much wind and solar-based electricity on the other power
plants in the 50Hertz TSO area.

However, the power supply from the solar panels and wind turbines covered
a substantial proportion of the demand. The red line in Figure 3.33 shows that
the minimum load of the other power providers was just 4 GW while they had to
cover two substantial peaks. This substantially reduced the steady load, which has
negative consequences for nuclear and coal-based power plants. The traditionally
rather profitable daytime power supply almost halved for the power stations. For
the whole day, the renewable energy sources covered some 30% of the electric
energy demand.
Figure 3.34 shows an even larger impact that the power supply from wind turbines
and solar panels has on the output from the power plants. On May 12, 2012, the two
renewable sources covered almost 70% of the total electricity demand. Between 2 pm
and 5 pm, the power plants needed to supply only 1 GW to the grid. Again, if just two
500 MW power plants would then have to fill the demand not covered by the renew-
ables, a tripping of one of them would mean losing so much dispatchable capacity that
primary reserves could never fill the gap.
The impact of the solar PV panels and wind turbines on electricity production
in the German 50 Hertz TSO area shows that the typical baseload often disappears.
The need for at least a certain amount of dispatchable power in the system means that
occasionally electricity has to be exported to neighbouring areas. Even the curtailment
of wind turbine output is needed now and then to avoid an uncontrollable situation,
such as there being limited export opportunities through transmission restrictions or
insufficient demand.
72  Power supply challenges

3.5. Effects of photovoltaics on other power plants in Texas and


California
The Texas transmission operator ERCOT provides hourly data on the daily electricity
demand in its region. CAISO from California also provides the output from the dif-
ferent electricity sources. This opens up opportunities for discussing the effect of pho-
tovoltaics on fuel-based generators.
Figure 3.35 illustrates how power demand in the ERCOT region varied during
2012. In contrast to Finland, Ireland and Germany, the demand for power is maxi-
mised during the summer. This is because of the high ambient temperatures that
require a substantial amount of cooling for households, commercial buildings and
factories. The variations in demand during the daytime are apparently much higher
in the summer. The power demand can vary between 35 GW and 65 GW on a hot
day. Figure 3.36 shows power demand plotted against the average daily ambient

80 000 40

temperature in Texas (°C)


ERCOT power demand

60 000 30

Average daily
(MW)

40 000 20

20 000 10

0 0
0 2 000 4 000 6 000 8 000
January – December hours in 2012

Figure 3.35.  Hourly data of power demand in the ERCOT region during 2012.

80 000
ERCOT power demand

60 000
(MW)

40 000

20 000

0
10 15 30 25 30
Average daily temperature in Texas (°C)

Figure 3.36.  Power demand in the ERCOT region plotted against the daily average
temperature in Texas (year 2012).
3.  Balancing power demand and supply when conditions change  73

temperature in Texas. The correlation


between the temperature and demand is

ERCOT power demand (MW)


70 000
obvious. For relatively low temperatures, 60 000
electricity demand in the state also 50 000
tends to increase. The overhauling of P > 40 GW only 29% of the time
40 000
power plants normally takes place in 30 000
the winter when the electricity demand 20 000
is much lower. Occasionally, however, 10 000
during extremely cold spells, the power 0
demand can reach 50 GW. In the past 0 10 20 30 40 50 60 70 80 90 100
rolling black-outs have occurred because % of the time, year 2012
of insufficient capacity, also caused by
frozen cooling systems. Figure 3.37.  Distribution curve of power demand in the
Although peak electricity demand ERCOT TSO region for 2012.
in the ERCOT region approached 70
GW in 2012, demands higher than 40 GW occur for less than 2500 hours per year
(figure 3.37). This means that a large fraction of the generating capacity in Texas will
have a low utilisation factor. This requires generating equipment with a relatively
low capital investment per kW. Due to the variability in demand, these power plants
should be able to experience many starts
and stops. In addition, they should have August 7, 2012, Texas
high power output ramping up and down 70 45
Power demand (GW)

Power demand (GW)


rates.

Temperature (°C)
60 40
Figure 3.38 gives an example of the 50 35
pattern in power demand and ambient
40 30
temperature during a hot day in August Austin temperature °C
30 25
2012. During the night, less than 40 GW
20 20
is required, while at around 4 pm when
0 4 8 12 16 20 24
the temperature reaches its peak, some Time(hours)
65 GW is needed to satisfy demand.
The maximum ramp up rate is 8.5 GW Figure 3.38.  Hourly power demand in the ERCOT region and
per hour. Figure 3.39 clearly illustrates the hourly temperature in Austin, Texas on August 7, 2012.
that there is almost no time lag between
power demand and temperature. Apparently, the insulation of buildings in Texas is
poor. Power demand could be substantially reduced during high and very low tem-
peratures with better insulation.
The direct relationship between power demand and ambient temperature might
offer interesting opportunities for saving fuel consumption with solar photovoltaics.
Figure 3.39 gives simulated results for a case where the maximum output from solar
PV equals 10 GW and a case with a peak PV output of 20 GW. Solar PV clearly
provides smoothing of the power plant output during the daytime. However, as in
Germany, a high ramping up of power plants is needed where there is a substantial
amount of PV capacity when the sun sets. In addition, the utilisation factor of the
74  Power supply challenges

peaking plants will be further reduced


by PV capacity. This situation clearly 70
Power demand
requires generating equipment that can
60
quickly start and rapidly ramp up in
output. In addition, as in the case of the Case 1 generator output
50
other areas discussed in this book, the Case 2 generator

Power (GW)
output
amount of installed power plant capacity 40

cannot be reduced as a result of the


30
introduction of solar photovoltaics. Nev- Case 2: Max 20 GW
ertheless, a PV system with a maximum of solar PV output
20
output of 20 GW would have produced
113 GWh in Texas on August 7, 2012. 10
Case 1: Max 10 GW
That would have saved some 900 TJ of of solar PV output
0
fuel on a single day for a power plant
0 4 8 12 16 20 24
efficiency of 45%. The 900 TJ equals 25 Time of the day (hours)
million standard m3 of natural gas with a
lower calorific value of 36 MJ/ m3. This
Figure 3.39.  Simulated cases where photovoltaic capacity
is enough fuel for one million passenger has been added in the ERCOT region.
cars to drive at least 300 km each.
The Californian transmission system operator CAISO makes available hourly
data on power demand and generation. Generation is split into the different elec-

Figure 3.40.  Although wind and solar energy compensate each other to some extent,
back-up capacity is always needed.
3.  Balancing power demand and supply when conditions change  75

tricity sources, such as those based


on fuel, wind and photovoltaics. This California, October 16, 2013
makes it easy to determine the effect of 18 000
16 000
renewable energy on the output from
14 000
fuel-based power plants. The data of the
12 000

Power (MW)
solar PV output in figure 3.41 show that Thermal power plants
10 000
the maximum output of PV panels was Solar PV outputs
8 000
2 GW on October 16, 2013. The light 6 000
Thermal if doubling of PV
blue curve representing the output from Thermal if no PV
4 000
the thermal power plants is quite flat 2 000
during the daytime. A relatively small 0
peak in the output of the thermal power 0 4 8 12 16 20 24

plants occurs in early evening when the Time of the day (hours)
sun sets. If the installed PV capacity was
doubled, the green curve would apply Figure 3.41.  The effect of the output of solar PV panels on
for the thermal power plants. The green the output of thermal power plants in California on October
curve clearly tends to peak more than the 16, 2013.
light blue and the dark blue curves. This
requires flexible backup generators. Doubling the solar panel capacity in California
would have saved the equivalent of 3.6 million m3 of natural gas on October 16,
2013.

3.6. Conclusions
Matching power demand and supply becomes more complicated with variable and non-
dispatchable renewable energy sources. In areas where wind and solar energy have an
installed capacity close to or higher than half of the average power demand, baseload
generally disappears. The capacity factor of thermal power plants decreases – but their
installed capacity has to be almost the same as when there are no renewables in the
system.
What is obviously needed is flexible backup power. Backup power plants should be
capable of rapid, continuous starting and stopping. They should ramp output quickly
up and down and they should be fuel-efficient at any load. Where hydropower is not
present, the best solution appears to be gas-fired plants based on multiple combustion
engines. They can act as reserve capacity, answer to wind forecasting errors and follow
the output of renewables as closely as possible. In other words, agile gas-based power
plants are able to match power demand and supply once again.
4 Active and
reactive power
Electrical engineers distinguish between active and reactive power in electricity
supply systems. Active power refers to the energy transfer from an electric
generator to a load via conducting wires. Reactive power results from currents
created by alternating voltage in certain elements of the system without a net
release of energy. Reactive power affects the transmission capability of high-voltage
lines and the voltage in the system. Large shares of renewable energy sources, long
transmission lines, and reliance on a few large power plants can cause a lack in the
reactive power supply and problems with the reliability of the supply system.
Note from the author: The relationship between an increasing amount of
renewable energy in a system and problems with reactive power is not well
covered in available publications. This is why these issues are addressed here, even
though the content becomes rather technical. Students, researchers, consultants,
business managers, and policy makers should benefit from this chapter. Readers
not interested in technical details can jump to sections 4.6. and 4.7. where
the consequences of renewable electricity for the reactive power supply are
summarized.
78  Power supply challenges

4.1. Reactive power analogies


Reactive power is like the foam on top of a glass of beer, where the liquid in the glass
represents active power. The foam takes space in the glass and therefore reduces the
beer-containing capacity. By comparison, reactive power reduces the electricity trans-
port capacity of a power transmission line. Another simplified analogy for active and
reactive power is a train. The passengers carried by the train can be compared with
active power, while the unavoidable train’s mass and volume can be seen as a burden,
like reactive power. Reactive power results from current in the system that transfers no
energy. Reactive power affects not only the active power capacity of transmission lines,
but also voltages in the system.

Figure 1.1.  Analogies for active and reactive power.

4.2. The three basic load elements in alternating current systems


R eactive power and its consequences for a power supply system can only be properly
understood with some basic knowledge of electrical systems. In the following sections,
the three basic elements in electric circuits will be explained. These elements are the
resistors, the capacitors and the inductors. The intention is not to repeat any electrical
engineering textbook, but to use illustrations together with some basic mathematics to
provide an introduction to the nature of active and reactive power.

4.2.1. Resistive load


Active power results when a voltage is applied to a resistive load. A resistive load R is an
element that passes an electric current without any delay when a voltage is applied to it,
while it turns electric energy into heat or work. One example of a pure resistor is an elec-
4.  Active and reactive power  79

tric heater. The current IR through the resistor is directly propor-


tional to the applied voltage VR. The resistance R is expressed in
the unit ohm (Ώ):
VR
R= Equation 4.1.
IR
The power P released in the resistive load is the product of
voltage V and current I. The unit of voltage is volt (V) and the
unit of current is ampere (A). The unit of power, or energy
flow, is watt (W), equalling joule/second (J/s).
P =V · I Equation 4.2.
Figure 4.2.  An electric heater,
If the voltage applied to a resistor is a sinusoidal wave, the an example of a resistive load.
resulting current through the resistor is also a sine wave. A
sinusoidal current is basically an alternating current (ac). Figure IR
4.3 gives a representation of a sine-wave voltage generator with
a resistor as the load. The current IR is a result of the voltage VR .

~
Figure 4.4 is an illustration where a sinusoidal voltage with
Generator
a peak value of 325 volt is applied to a resistor of 2.5 ohm. The R VR
peak value of a sine wave is also called amplitude. The ampli-
tude expresses the maximum excursion of the sine wave from
zero. Since the current through a resistive load follows the
voltage immediately, a sinusoidal current with a peak value of Figure 4.3.  A schematic repre-
325 V/ 2.5 Ώ = 130 A results (the black curve in Figure 4.4). sentation of an ac generator with a
resistive load.
Since both the current and voltage waves in the example have
their peaks and valleys at the same time, they are referred to as
being in phase.
The product of the instantaneous values of voltage and current gives the power
dissipated in the resistor. This product is illustrated by the solid red line with a peak
value of 325 V · 130 A = 42250 watt ≈ 42.3 kW. The voltage and the current oscil-
late around an average value of zero; half of the time they are positive and half of
the time they are negative. The resulting solid red power curve appears always to be
above zero. The reason for this is simple: basic mathematics tells us that multiplying
two negative values results in a positive value. The power curve oscillates around an
average value of 21.1 kW and apparently has the double frequency of the voltage. The
offset of the power curve from the horizontal axis (the dashed dark red line) equals
the effective averaged power dissipated in the resistor.
It is interesting to see from the solid red line in Figure 4.4 that the instanta-
neous energy delivered to a resistor is never constant in an alternating current
system. This generally does not affect the user because of the high frequency. The
light coming from an incandescent lamp will follow the oscillations in power to
some extent, but the human eye is hardly sensitive to light variations with frequen-
cies higher than 50 Hz. If the periodic variations in voltage level or power are a
80  Power supply challenges

problem for the application, a rectifier is


used to transfer the alternating voltage
into direct voltage. A direct voltage 1 000 50 000
does not have the periodical variations 800 40 000
of alternating voltage. Computers and

Voltage V and current I


audio equipment are examples of elec- 600 30 000

tric appliances that use direct voltage. 400

Power P
20 000
We will now mathematically deter-
200 10 000
mine the average power dissipated in a
resistive load. This will also reveal the 0 0
concept of the effective value or root –200 –10 000
mean square value (rms) of an alter-
–400 –20 000
nating voltage or current. The voltage
0 10 20 30 40 50 60
and current values of alternating elec-
Time (millisecond)
tricity systems are normally always given
as rms values. Voltage Current Instaneous power
Effective power
The result of multiplying two sine
waves can be found by using the basic
mathematical relationship:
Figure 4.4.  A sinusoidal voltage with a peak value of 325 volt
sin x · sin x = sin2x = ½ – ½ cos 2x , inducing a sinusoidal current with a peak value of 130 ampere
in a resistor of 2.5 ohm (Ώ), resulting in a cosine-shape power
Equation 4.3.
curve around a positive offset of 21.2 kW equalling the effec-
tive power delivered.
We can now give each sine wave an
amplitude of A and B, so that the equation becomes:

A sin x · B sin x = A B – ½ AB cos 2x Equation 4.4.


√2 √2
This shows mathematically that the product of two in-phase sine waves results
in a constant equalling the product of the amplitude (= peak value) divided by √2 of
each sine wave, and in a varying component.
If we use the amplitude values of our example of 325 V for voltage and of 130 A
for current, the constant turns into 325/√2 · 130/√2 = 21 125. This is exactly the
value of the effective power as shown by the dashed line in figure 4.4. The variable
component in the product of two sine waves clearly has double the frequency of the
initial voltage wave. This is indicated by the 2x in the cosine function.
Since the average dissipated power in the resistive load equals the product of the
voltage amplitude divided by √2 times the current amplitude divided by √2, we call
the peak value of a sine wave divided by √2 its ‘effective’ value or ‘root mean square’
(rms) value:
Vrms =Vpeak /√2 Equation 4.5.
4.  Active and reactive power  81

In the case of a sine wave with a peak value of 325 volt, the effective value Vrms
= 325/√2 = 230 volt. This is a familiar number for many people. Again, in electrical
engineering, what is always meant when talking about voltage and current, unless
otherwise stated, is the rms value of an alternating voltage or current. Since the
rms value of the current passing a resistor of 2.5 Ώ for an rms voltage of 230 volt is
230/2.5 = 92 ampere. The power P dissipating in this resistor equals then:
P = Vrms . Irms = 230 . 92 = 21125 W ≈ 21.1 kW Equation 4.6.

This resulting average product of voltage and current is called active power.
The visualisation of the product of the two sine waves, as given in figure 4.4, is
intended to help the reader to understand better the result of the dry mathematical
manipulation. The reader should now be able to understand the concept of active
power, where voltage and current are in phase and their rms values determine the
power dissipated in a resistive load.

4.2.2. Capacitive load


Capacitors are electrical components that store an electric charge on plates or conduc-
tors that are separated by a non-conductive layer, the so-called dielectricum. This dielec-
tricum can be air, as is the case for power transmission lines in parallel. Unlike resistors,
capacitors do not dissipate energy.
The relationship between the voltage V between the plates and the electric charge
Q on one of the plates is given by the capacitance C:
Q
C= Equation 4.7.
V
Without a difference in charge between the plates, there will be no voltage between
the plates. A capacitor can be compared to a water tank; a stream of water has to fill
the tank in order to create the water level. In order to create a voltage over a capacitor,
an electric current has to charge a plate of that capacitor. This means that the current
has to be earlier than the voltage. In other words, the current leads
the voltage. The capacity C is expressed in farad (F) and the charge – – – – – – – – – –
is expressed in coulomb (C). In alternating current systems with
only a capacitor connected to the generator, the current will lead a
sinusoidal voltage by 90 degrees. The relationship between voltage Dielectrium
and current is the reactance XC of the capacitor:
VC 1
= XC = Equation 4.8. + + + + + + + + + +
IC j2πfC
Figure 4.5.  A basic representa-
in which ƒ is the frequency of the voltage wave in hertz. The j is tion of a capacitor. In ac systems,
the charge on the plates oscil-
a so-called operator indicating the 90° phase shift between voltage
lates between positive and nega-
and current. The reactance of a capacitor appears to be inversely tive.
proportional to the voltage frequency, meaning that the reactance
82  Power supply challenges

of a capacitor with a given capacitance


C diminishes with increasing frequency. 500 25 000

Voltage V and current I


400 20 000
Figure 4.6 shows the voltage, current, and 300 15 000

Product V × I
the product of voltage and current for a 200 10 000
100 5 000
peak voltage of 325 volt and a capacitance 0 0
–100 –5 000
of 1500 μF for a frequency of 50 Hz. The –200 –10 000
product of voltage and current has again –300 –15 000
–400 –20 000
double the frequency of the voltage, but it –500 –25 000
0 10 20 30 40 50 60
has on average no offset from zero. There-
Time (millisecond)
fore, although the current reaches peaks of
Voltage Current Voltage × current
153 A, on average no power is delivered
Average product
to the capacitor, as indicated by the green
dashed line. The product of voltage and
current is, therefore, called reactive power, Figure 4.6.  Reactive power as the product of voltage and
with the unit volt ampere (VAr). current in the case of a purely capacitive load
The natural capacitance in most
electricity supply systems is relatively small. Underground electric cables have more
capacitance than overhead cables due to the closer proximity of the conductive wires.
Also, some modern electronic devices have a capacitive element next to a resistive
one. Such an example is the LED light.
We will now introduce the phasor or vector concept that is commonly used
in electrical engineering. A sine wave is actually a projection of the position of a
point moving along a circle at a constant speed. Point A in Figure 4.7 starts at 0°
and moves in an anti-clockwise direction along a circle with a radius r. The vertical
distance of point A from the starting point is plotted against a vertical axis, while
the angle covered from the starting point is plotted against a horizontal axis. The
amplitude, i.e. the maximum excursion from the starting point, equals the radius
of the circle. The dark green arrow in the circle is called the phasor of point A. If
point A arrives at 90°, the phasor points in a vertical direction. If the phasor covers
the full 360° of the circle in 20 milliseconds, the so-called period time of the sine
wave equals 20 ms. A period time of 20
ms means that point A completes 50 90°
revolutions in one second. The frequency
of the sine wave is in that case 50 hertz Amplitude
(Hz). r
180°
Figure 4.8 gives the symbolic rep- A

resentation of a generator with a purely


capacitive load. As mentioned earlier,
the current leads the voltage over the 270°
0 30 60 90 120 150 180 210 240 270 300 330 360
capacitor by 90°. This is indicated by Degrees
the vertical current phasor and the hori-
zontal voltage phasor on the right-hand Figure 4.7.  The sine wave as the projection of a point (A),
side of the illustration. moving anti-clockwise along a circle
4.  Active and reactive power  83

To summarize, because of a capaci- IC


tive element in an AC electricity supply
system, the voltage and current are no IC
longer in phase. For a purely capacitive
element, the current wave leads the voltage ~ VC
90°
VC
wave by 90 degrees. On average, no energy
is dissipated in a capacitive element.
Figure 4.8.  Representation of a generator with capaci-
tive load, with the voltage phasor and the current phasor
4.2.3. Inductive load leading the voltage by an angle of 90 degrees

Inductors consist of coils of wound conduc-


tive wire that create a magnetic field when an electric current flows through them. This
magnetic field attracts iron. Electric motors derive their motive power from this prop-
erty. An inductor can be compared to a car that needs a driving force in order to accel-
erate. To give motion to the mass of a car, a force has to be applied to the mass before
speed results. The speed of the car, therefore, always lags behind the driving force. By
comparison, the current through a pure inductor lags behind the driving voltage. If the
voltage across the inductor is a sine wave, the current through the inductor lags behind
the voltage by 90° resulting in a 180° shifted cosine wave. The 90° lagging equals 5.0 ms
in a 50 Hz system and 4.17 ms in a 60 Hz system.
Many elements in an electricity supply system, such as electric motors, fluores-
cent lights and transformers, have inductance. Even a simple straight wire has some
inductance and thus long overhead transmission lines can have a significant induc-
tance. Figure 4.10 represents a generator with a purely inductive load. The phasor
diagram shows that the current is lagging the voltage by 90°. Figure 4.9. 
The relationship between voltage V L and current I L for an inductor with an An electric coil.
inductance L equals:
VL Equation 4.9.
= XL = j2πfL
IL
in which L is the inductance expressed in henri (H). The j again indicates a 90°
phase shift between the voltage and current, while f is the frequency of the voltage.
The reactance of an inductor is directly
proportional to the frequency, meaning IL
that the reactance of a given inductance
increases with increasing frequency.
Figure 4.11 gives the voltage and cur-
rent, as well as their product, in the case
~ VL 90° VL
of an inductive load, where the yellow IL
curve is voltage and the blue curve is the
resulting current. The averaged product Figure 4.10.  Diagram of a generator with a purely induc-
of voltage and current, the dashed purple tive load and the voltage phasor with the lagging current
phasor.
curve, is zero as in the case of a capacitor.
84  Power supply challenges

Nevertheless, the instantaneous value


500 25 000
of the product of voltage and current
400 20 000
reaches high values. The product of

Voltage V and current I


300 15 000
voltage and current has again double 200 10 000

Product V × I
the frequency of that of the basic voltage 100 5 000
0 0
wave. In the example of figure 4.11, the
–100 –5 000
peak in current reaches 130 A for a peak –200 –10 000
voltage of 325 V and an inductance of –300 –15 000
8 millihenri (mH) for a frequency of –400 –20 000
–500 –25 000
50 Hz. The product of voltage and cur-
0 10 20 30 40 50 60
rent is again the reactive power.
Time (millisecond)
Voltage Current Voltage × current
4.3. The power factor cos φ Average product

This section will explain the power factor


cos φ. Knowledge of the background of the Figure 4.11.  Reactive power as the product of voltage and
power factor is important in understanding current in the case of a 50 Hz voltage with a peak value of
its effect on reactive power demand in 325 V acting upon a purely inductive load of an inductance of
8 millihenri.
power supply systems and on maintaining
the proper voltage level. Real electricity consuming elements, such as electric motors,
ovens and lights, can have resistance, inductance and capacitance at the same time.
The electric motor shown in Figure 4.12 is a typical example of a device with
resistance and inductance. Figure 4.13 is a schematic representation of such a device.
The current IZ flows though the inductor as well as through the resistor, since these
two elements are in series. This current is plotted in Figure 4.13 as an arrow, a
phasor, along the horizontal axis. The previous section has shown that the voltage
over an inductance leads the current by 90°. Therefore, we have to plot the voltage
over the inductive element as shown by the green VL
phasor. The voltage VR over the resistor is in phase with
the current, so it is plotted along the horizontal axis. The
resulting voltage across the combination of resistor and
inductor equals:

Equation
VZ = √(VL2 + VR2) = IZ √((2πƒL)2 + R2)
4.10.
Figure 4.12.  An electric
motor has resistive and The quotient VZ/IZ is called the apparent resistance
inductive elements
or impedance Z. The word impedance has been derived
from the Latin impedio, meaning hindrance. Literally, it
means that one’s feet, the pedes, are wrapped. In the example of Figure 4.13, the rms
voltage VZ equals 230 V. Since the impedance Z equals √ ((2π · 50 · 0.04)2 + 202) =
23.65 Ώ, the resulting current IZ equals 230 /23.65 = 9.73 A. The voltage V Z leads
the current IZ by the angle φ. The angle φ can easily be found from cosine φ = V R
/ VZ . In this case, cosine φ equals 194/230 = 0.84, which is the power factor cos φ.
4.  Active and reactive power  85

The angle φ equals 32.9° in this example, V L

since cos 32.9 = 0.84. The power factor I Z


V
122 V V
230 V L
Z

indicates which fraction of the voltage Impedance Z L = 40 mH


over the impedance is actually available φ V I R Z
V = 230 V Z R = 20Ω V
for the resistor for dissipating energy.
R
194 V 9.73 A

Knowledge of the power factor is indis- ƒ = 50 Hz

pensible when designing and operating Power factor cos φ =


Z = √((2πƒL) + R ) = 23.65 Ω 2 2
194/230 = 0.84
power supply systems. I = V /Z = 230/23.65 = 9.73 A
Z 2

The electrical load, consisting of an


inductance L and a resistance R, can be Figure 4.13.  Example of an electrical load with inductance
replaced by the impedance Z (Figure 4.14). and resistance, resulting in a phase shift between voltage and
The phasor diagram can be redrawn with current.
the voltage VZ at the horizontal axis and
the current IZ lagging behind the voltage by an angle φ. The current can then be split
into a phasor Iactive = IZ  cos φ and a phasor Ireactive = IZ sin φ. Only the real part of the
current will contribute to energy release in the impedance. Therefore, the active power
to the impedance equals:
IZ
Pactive = VZ · IZ cos φ Equation 4.11.
Iactive = 8.17 A VZ = 230 V

~
and the reactive ‘power’ to the impedance φ
equals: VZ Z
IZ = 9.73 A
Generator Ireactive = 5.28 A
Preactive = VZ · IZ sin φ Equation 4.12.

The unit of active power is watt (W) Figure 4.14.  Electric current to an electrical impedance split
and the unit of reactive power is VAr. into a an active-real-component and a reactive – imaginary –
component.
In the examples shown in figures 4.13
and 4.14, the active power is 230 · 8.17
= 18.8 kW and the reactive power is
400 20
230 · 5.28 = 12.1 VAr. Due to the reac- Voltage (V)
300 15
tive part of the impedance, the electric Current
200 10
current from the generator to the active
Voltage VZ

Current IZ
100 5
load is a factor of 1/cos φ, or 9.73/8.17 0 0
= 1.19 here, higher than in the case of –100 –5
no reactive part. It means that both the –200 –10
generator and the transmission line have –300 –15
to be able to accept a higher current than –400 –20
the active current delivered to the load. 0 5 10 15 20 25 30 35 40 45 50
The blue curve in Figure 4.15 gives Time (millisecond)
the resulting current for an rms voltage
of 230 V for an impedance of 23.65 Ώ Figure 4.15.  Diagram of voltage and current for Vrms = 230
and a power factor cos φ of 0.84 as per V and Irms = 9.73 A for an impedance of 23.65 Ώ and a power
factor cos φ = 0.84
the examples in figures 4.13 and 4.14. The
86  Power supply challenges

blue current curve follows the yellow voltage curve by the angle φ, equalling 32.9°
here, since cos 32.9 = 0.84. For a 50 Hz system, an angle of 32.9° results in a time
shift between voltage and current of 32.9/360 · 20 ms = 1.83 ms, since a complete
50 Hz sine curve of 360° covers 20 ms.
To summarize, the power factor in an ac electrical system indicates the phase
shift between the voltage wave and the current wave. In systems with a power factor
lower than 1, voltage and current are no longer in phase. In addition, the electric cur-
rent in a system with a power factor lower than 1 is always higher than in the case of
a power factor of 1, provided the supply voltage is the same.

4.4. Impedance of electricity transmission systems


When a generator supplies both active and reactive power to its load, the transmission
system between the generator and the load has to transport both components of the cur-
rent. Transmission and distribution lines also have their own electrical impedance. The
generator ‘feels’ the sum of the transmission system impedance and the load impedance.
Since reactive power increases the total current in the power lines, more energy will be dis-
sipated in the resistance of the power lines, and this results in a further reduction in voltage
over the line resistance. In addition, the reactance of a power transmission line itself in
combination with a highly reactive load can cause a substantial voltage drop over the line.

4.4.1. Transmission line resistance


The electrical resistance of a transmission line depends upon the length of the cable, its
diameter, its material properties and the temperature. Copper has a low electrical resis-
tance, but it lacks the tensile strength of steel. Tensile strength is needed when wires are
bridged between transmission towers. The mechanical load on the wires will increase in
case of gales or when freezing rain causes ice to form on the wires. Copper is relatively
expensive and therefore aluminium is often used as a conductor. Again, aluminium lacks
the needed mechanical strength. That is why a modern transmission line consists of an
inner steel cable with aluminium conductors wrapped around it. Aluminium has roughly
double the tensile strength of copper, while steel is twice as strong as aluminium.
The actual electrical resistance of a conductor can be found by multiplying the
resistivity ρ by the length l of the conductor and dividing it by the area A of the con-
ductor:
ρl
R= Equation 4.13.
A
For a conductor diameter d of 25 mm, equalling an area A of 491 mm2, an alu-
minium transmission line has a specific resistance of about 2.82 · 10 –8/ (491 · 10 –6)
= 5.74 · 10 –5 Ώ per meter. This seems to be a small number, but for a transmission
line of 250 km the resistance of each of the three wires of a three-phase system is
250 · 103  · 5.74  · 10 –5 = 14.3 Ώ. For a current of 500 A per wire, such a resistance
4.  Active and reactive power  87

results in a power loss of I2 R of 5002  · 14.3


= 3.57 MW per wire, or 10.7 MW for the
total three-wire system. If the system
capacity is 500 MW, one loses slightly
more than 2% of the energy over a dis-
tance of 250 km. Should the transmission
line have only resistance and no reactive
components, the voltage loss would be
I · R = 500 · 14.3 = 7.15 kV. For a 380 kV
line, such a voltage loss is relatively small.
In reality, transmission-tower-based power
lines have considerable inductance that Figure 4.16.  Two double three-phase high-voltage transmis-
substantially affects the voltage drop. This sion lines with transmission towers carrying the cables
will be shown in section 4.4.3.
First, we have to check if the energy loss in the transmission line will not over-
heat the wires. A higher temperature will increase the resistivity ρ and reduce the
mechanical strength of the wires. Generally, an increase ∆T in the operating tem-
perature of 50 K with respect to 20 °C, i.e. an operating temperature of 70 °C is
considered as being the limit. As a rule of thumb, an overhead aluminium cable with
its wire area A expressed in mm2 can accept a current of:
Imax = 1.14 · A¾ · √ΔT Equation 4.14.

In the example above, with a conductor area of 491 mm2, the maximum allowed
current equals 838 A. The presumed current of 500 A is therefore not a problem
when the ambient temperature is low. The resistivity itself depends on the tempera-
ture:
ρactual = ρ0 [1 + α (Tactual – T0)] Equation 4.15.

in which α is the temperature coefficient of the material. For aluminium, α equals


0.0039/K. This means that a temperature rise of 50 K increases the resistance of a
power line made from aluminium by almost 20%. This factor has to be considered in
hot areas when electricity demand peaks because of air conditioning.

4.3.2. Transmission line inductance


Table 4.1.  The electrical resistivity ρ of
The inductance of a set of wires transmitting electric energy some conductive materials.
is relatively low per meter length. However, ac transmission
Material Resistivity ρ at 20 °C
lines can cover distances of up to a thousand kilometres. (Ώ m)
Then the inductance plays an important role in the energy
Copper 1.70  10 –8
transport capacity of the transmission system. The specific
inductance of a high-voltage transmission wire depends on Aluminium 2.82  10 –8
its inner radius r and the distance s from the other wires:
Steel 14.3  10 –8
88  Power supply challenges

μ
Lspecific = π (¼ + ln rs ) Equation 4.16.

in which μ is the magnetic permeability in a free space, equalling 4π · 10–7. For s = 2 m


and r = 12.5 mm, Lspecific equals 2.13 μH/m. Theoretically, free space means a vacuum,
but ambient air as a medium around the wires does not make much difference.
For a line frequency of 50 Hz, a specific inductance of 213 μH/m renders a spe-
cific inductive reactance X l (spec) of:
Xl(spec) = j2 · π · 50 · 2.13 = j 670 μΩ/m Equation 4.17.

4.4.3. Transmission line capacitance


An overhead power transmission line has a relatively low capacitance compared to
ground cables. Yet, its capacitance cannot be neglected since the current needed to
charge and discharge the wires has to be provided by the generator and transmitted
through the wires. In electrical engineering, the capacitance per meter of wire length of
a single phase system is approached by:
πε
Cspecific = s Equation 4.18.
ln r

Here, ε is the permittivity, which has a free space value of 8.85 pF/m (p = pico = 10 –12).
Again, ambient air can be considered as free space here. The distance between the wires
is s again and the radius of the wire is r. A larger distance between the wires reduces
the capacitance, while a larger wire diameter increases the capacitance. Using the same
line dimensions as for the inductance, we find that Cspecific = π · 8.85/ln (2/0.0125 ) =
5.48 pF/m. This results in a specific capacitive reactance XC(spec) = 1/j2πƒC = –j 581
MΏ/m (M = 106 and the product of the operators j · j is by definition –1).

4.4.4. Transmission line impedance


In principle, a power line can be represented by a
Lspec Rspec
chain of multiple small sections of, say, one meter
length having some resistance, inductance and
capacitance. Actually, some conductivity between
the lines and the earth should also be taken into Cspec Gspec
account since electrical charge tends to leak away
via insulators and the air. Especially in misty
weather, the typical crackling noise caused by the
discharging can be heard. Figure 4.17 gives a repre- Figure 4.17.  A diagram of a small section of a
sentation of a small section of a transmission line. power transmission system with resistance Rspec,
inductance Lspec, capacitance Cspec and shunt resist-
The shunt resistance Gspec is a measure of the con-
ance Gspec
ductivity between the line and the earth.
4.  Active and reactive power  89

Determining the total impedance of a power transmission line consisting of a


large number of infinitesimal sections as shown in Figure 4.17, requires advanced
mathematical knowledge. Fortunately, the impedance of a transmission line of
medium length l , up to 250 km, can be approached by a so-called π section as
shown in figure 4.18.

4.5. Voltage change over a power transmission line


This chapter aims to explain how the voltage at the end of a transmission line is affected
by the impedance of the line and by the nature of the load. Section 4.5.3 will reveal how
the nature of the load affects the power carrying capacity of the transmission line. As an
example, a power line of 200 km will be used, with a resistance of R spec of 50 μΏ/m, an
inductance of Lspec of 2 μH/m, a capacitance of Cspec
of 5 pF/m and a neglected shunt resistance Gspec. l · Lspec l · Rspec
With these data, the ratio of the voltage Vin
at the beginning of the line from the generator
and the voltage Vout at the end of the line can be Vin ½ · Cspec ½ · Cspec Gspec
determined. This can be done without a load, with
a purely resistive load, or with any impedance. In
this example, a single phase system will be used.
Although real AC transmission systems are three- Figure 4.18.  Diagram of a replacement for a
phase, the simplified example is used to show the power transmission line of up to 250 km.
mechanisms involved.

4.5.1. Case without load at the end of the transmission line


Normally, overhead high-voltage electricity transmission lines are disconnected from
the supply system when they do not carry any load. However, with a substantial amount
of varying renewable electricity sources in a system, disconnecting is less easy since the
power available via the transmission line can serve as a backup for the renewable gen-
erators. Moreover, indirectly coupled renewable energy sources do not often provide
reactive power. Therefore, it is interesting to see how the voltage Vout at the end of such
a transmission line differs from the voltage Vin at its beginning when there is no load.
This extreme case shows how even a non-loaded transmission line introduces a change
in voltage. Figure 4.19 shows the replacement scheme of a 200 km long transmission
line, again with the resistance of R spec of 50 μΏ/m, an inductance of Lspec of 2 μH/m and
a capacitance of Cspec of 5 pF/m.
Since there is no load at the end of the transmission line, current Iline passes
three elements of the line, i.e. the inductor of j 126 Ώ, the resistor of 10 Ώ and the
final capacitor of –j 6 366 Ώ. The net reactive part of the impedance Zline = Vin / Iline
equals –j  6 366 + j 126 = –j 6 240. Therefore, Zline = √ (102 + 6 2402) ≈ 6 240 Ώ.
This shows that the resistance of only 10 Ώ and the inductance of 126 Ώ have practi-
cally no effect on the impedance in this case, so that the impedance is dominated
90  Power supply challenges

by the capacitor. The line current Iline equals Vin j 126 Ω 10 Ω


lgenerator lline
/6 240, and therefore the voltage Vout equals 6 366
· Vin /6 240 = 1.013 Vin. The voltage at the end
of the transmission line is, therefore, higher than Vin – j 6366 Ω – j 6366 Ω Vout
the voltage at the beginning. The generator cur-
rent Igenerator is slightly more than two times higher
than the line current Iline in this case because of
the two impedances of –j 6 366 and –j 6 240 in Figure 4.19.  Model of a 200 km long single-
parallel. The generator current leads the generator phase transmission line.
voltage by almost 90° because of the capacitive
character of the load.
For readers without a background in electrical engineering, the fact that the
voltage at the end of a transmission line is higher than at the beginning might appear
as a miracle. However, a purely capacitive load on an overhead transmission line with
its typical inductive properties always increases the voltage at the end of such a line.
A raised voltage at the end of a transmission line can create problems for appliances
and local generators.

4.5.2. Case with a purely resistive load at the end of a transmission


line
The next example is a case whereby a 100 kV, 200 km long, transmission line is loaded
with a purely resistive load, as represented in figure 4.20. The voltage at the end of the
transmission line is shared by the resistance of the load and the shunt capacitor of the
transmission line of –j 6366 Ώ. In the case of a low resistance value Rload, the current
induced by the voltage Vout is dominated by the current through the resistor. Where the
value of Rload is high, the current approaches that of the case without any load.
If the resistance at the end of the power line is decreased to 10  Ώ, the line cur-
rent Iline reaches a value of 800 A (0.8 kA), which we presume to be the maximum
allowed current to avoid overheating. In Figure 4.21, the line current is shown as a
function of resistive load by the blue line. The yellow line shows that for this resis-
tive load of 8 Ώ, the voltage Vout at the end of the transmission line has dropped to
8% of the voltage Vin. This is way too high. A voltage drop to 90% of Vin is generally
considered as being ultimately accept-
able. A voltage drop to 90% is in this case j 126 Ω 10 Ω
lgenerator lline
reached for a load resistance of 281 Ώ. The
line current is then 321 A, while the power Vout
dissipated in the load equals 29 MW. For Vin – j 6366 Ω – j 6366 Ω
this load, the power loss in the transmis- Rload
sion line equals Iline2 · R line = 3302 · 10 = 1.09
MW, equalling some 3.8% of the power dis-
sipated in the load. The power factor cos φ Figure 4.20.  Model of a 200 km overhead transmis-
for the 29 MW load is 0.93 (inductive), as sion line with a purely resistive load.
4.  Active and reactive power  91

shown in figure 4.22. Therefore, for this


200 km transmission line the load limit
50 1.0

Power dissipated in load


caused by voltage drop is reached long

Vout /Vin ; I line (kA)


before the limiting line current of 800 A 40 0.8

reistor (MW)
is reached.
30 0.6
If the load resistance is increased
to 1 000 Ώ, the power dissipated in the 20 0.4
resistive load equals 10 MW. Figure 4.22 10 0.2
shows that for a load of 10 MW the power
factor of the line current Iline reaches 1. 0 0
0 200 400 600 800 1 000
The voltage Vout then roughly equals Vin.
Load resistance value Rload (Ω)
The line current is then 0.1 kA. For a
load resistance higher than 1000 Ώ, and
consequently a load lower than 10 MW, Figure 4.21.  Ratio of ingoing and outgoing voltage of a 100
the line current starts to lead Vin. Vout is kV transmission line, with the line current and the power dissi-
then slightly exceeding Vin as in the ear- pation in the load depending upon the resistance of the load.
lier example of no load at all, as can be seen from the yellow line in figure 4.21.
For a load resistance lower than 100 Ώ, the voltage over the load decreases so much
that the power dissipation in the load also decreases. The maximum power dissipated
in the load for the given transmission line dimensions and line voltage is, therefore,
37 MW with a current of 0.5 kA. Only shortening the transmission line and thereby
reducing its impedance will help in transporting more power to the load.

4.5.3. Case with a mixed load at the end of a transmission line


In reality, the load at the end of a transmission line is generally mixed, with a resistive and
an inductive part. Many renewable energy sources, such as wind turbines and solar panels,
provide only active power and do not par-
ticipate in providing reactive power, so that
the generators supplying the remaining 1.0
load will sense a highly reactive load. This
Cos φ line current lline

0.9
substantially lowers the capacity of a trans-
mission line between the load and the gen- 0.8
Capacitive Inductive
erators, as will be shown with an example. 0.7
Figure 4.22 shows the model of the power
supply system to be used for this. 0.6

In the case of a load consisting of 0.5


a resistor of 100 Ώ and an inductive 0 5 10 15 20 25 30 35
reactance of j100 Ώ, the impedance of Load (MW)
the capacitor at the end of the line is so
high compared with the load that its Figure 4.22.  Power factor cos φ of the line current Iline as a
contribution to the line current can be function of resistive load for the transmission system shown in
neglected. The impedance Zload of the figure 4.20.
92  Power supply challenges

load equals √ (R load2 + X load2) = √ (10000 + j 126 Ω 10 Ω


Igenerator Iline
10000) = 141 Ώ. The power factor cos φ of
the load equals R load/Zload = 0.71. This value Vout
Rload
of cos φ can be quite realistic locally when
there are many active-power-only providing Vin –j 6 366 Ω –j 6 366 Ω
renewables in the system. jXload
Because the capacitor at the load end of
the line has been neglected in this case, the
line impedance and the load impedance can Figure 4.23.  Model of a high-voltage 200 km over-
be presumed to share the same current Iline. head power transmission line with a load having resist-
ance and inductance
The impedance of the capacitor at the begin-
ning of the line is also quite high, so we will
also neglect that element. The combined inductive reactance of the system is then j126
+ j100 = j226 Ώ and the combined resistance 10 + 100 = 110 Ώ. This renders a com-
bined impedance of √ (1102 + 2262) = 251 Ώ. For a voltage Vin of 100 kV, the resulting
current Iline equals 100000/251 = 398 A. The voltage Vout over the load can then be
determined from Zload · Iline = 141 Ώ · 398 A= 56.12 kV. This represents a dramatic
voltage drop over the transmission line, since the voltage at the beginning of the line
is 100 kV. Yet, the power dissipated in the resistor of the load is only 3982 · 100 = 15.8
MW. This example illustrates that long high-voltage overhead transmission lines have
severe difficulty in transporting reactive power. Even for a relatively low active load,
the voltage at the end of the line easily collapses in the case of a load having a high
inductive component.
Figure 4.24 illustrates with phasors why the voltage over the load collapses so
easily in the case of a partly reactive load with a cos φ value of 0.7. The high line

Vin = 100 kV

VL(line) VR(line)
VL (line)
= 50 kV
Igenerator Iline j126Ω 10 Ω

100Ω
Vr(load)
VL (load)
Vin Vout = 40 kV Vout = 56 kV

VL(load) j100Ω φload


Vr(load) = 40 kV I line
Vr(line) = 4 kV

Figure 4.24.  Phasor diagram of a 200 km 100 kV transmission line with a load of
impedance Z of √ (1002 + 1002) = 251 Ώ and of cos φ = 0.7
4.  Active and reactive power  93

inductance compared to the load induc-


tance drastically reduces the voltage over 200 800
the load. The line current Iline lags behind 150
Vin I line
600
the voltage Vin at the beginning of the line 100 400
Vout

Voltage (kV)

Current (A)
by φline = 64°, since cos φline = (40+4)/100 50 200
= 0.44. 0 0
–50 –200
The voltage drop over the transmis-
–100 –400
sion line decreases if the values of the
–150 –600
resistor and the inductor of the load are
–200 –800
increased with respect to the values given 0 10 20 30 40
in figure 4.24. This automatically reduces Time (ms)
the power transmitted over the line. For
a value of R load as well as X load of 625 Ώ,
the voltage over the transmission line Figure 4.25.  Phase shift between the voltage Vin at the
drops by only 10%. Such a voltage drop beginning of the transmission line and the voltage Vout at the
is generally considered as the maximum end of the line, and the phase shift with the line current Iline for
the conditions shown in figure 4.24.
acceptable value. The power dissipated
in R load equals 6.5 MW in that case. By comparison, for a purely resistive load the
power dissipated in the resistor equals 26 MW for a voltage drop of 10%. Figure 4.26
illustrates that the energy transferring capacity of a high-voltage overhead transmis-
sion line heavily decreases when the load has a relatively low power factor.
In the example given, the maximum
amount of energy that can anyhow be
transferred for a cos φ of 0.7 is almost 16
Voltage ratio Vout/Vin (%)

100
MW (see Figure 4.26). If the impedance
cos φ = 1
of the load is further lowered, the power 90
dissipated decreases since the voltage over
80
the load further decreases. It will be clear cos φ = 0.7
that if the length of the line is shortened, 70
cos φ = 0.9
the voltage drop will be smaller. The line
inductance and resistance are directly 60
proportional to the length. 0 10 20 30 40
Active power of load (MW)

4.6. Risks created when


insufficient reactive power is Figure 4.26.  Ratio of ingoing and outgoing voltage over a
200 km 100 kV transmission line as a function of active power
supplied by renewable energy demand for different power factors cos φ of the load.
sources
This chapter has explained that power supply systems have to provide both active and
reactive power. Electric loads have an apparent resistance, or impedance. The imped-
ance consists of a resistive (real) part and a reactive (imaginary) part. The reactive parts
originate from the capacitive and inductive elements in the system. The same applies for
transmission and distribution lines. High-voltage overhead transmission lines have a net
94  Power supply challenges

inductance, while underground distribution cables have a net capacitance. On average,


the combined electric loads in a power supply system have a slightly inductive character
with a power factor close to 0.85. That means that the real part of the apparent resistance
or impedance has a value that is 85% of the impedance, while the inductive part equals
53% of the impedance. The square root of the sum of the real part squared and the
reactive part squared makes 100% again: √ (852 + 532) = 100. This has been explained
in more detail in the previous sections. Overhead transmission lines have difficulty in
transmitting reactive power of an inductive nature. Reactive power tends to drastically
decrease the voltage over an overhead transmission line.
Traditionally, where the electricity supply has come from power plants only, the
plants were built in close proximity to the locations where the energy was consumed.
This meant that the transmission lines were quite short. Therefore, voltage reduction
and energy losses over the transmission lines were small. In the case of remote loca-
tions, transformers with a controllable transfer ratio could adjust the voltage of the
end users to the desired nominal value. Large consumers with loads having a low
inductive power factor could be asked to install adjustable capacitor units. Capacitors
help to compensate for the reactive part of the load. Also, grid operators take mea-
sures to compensate for reactive power. So-called SVCs (static VAr compensator)
and Statcoms (static synchronous compensators) can provide fast-acting reactive
power for high-voltage electricity transmission networks. Sometimes, grid operators
install so-called rotating synchronous condensers. Such machines are synchronous
generators running without a prime mover, such as an engine or turbine, to drive
them. By adjusting the magnetic field of the spinning generator, a capacitive load, or
even an inductive load, can be created. Certain transformer types can also compen-
sate for highly inductive loads. Nevertheless, large power stations carried the bulk of
the reactive power in the past.
Currently however, with the advent of many local generators based on renewable
energy sources, such as solar radiation and wind, the relative number of active power
stations is often drastically reduced. Chapter 3 shows that even with an average elec-
tricity supply of 20% from wind turbines, occasional high winds can push almost all
power stations from the grid. The same applies for areas where more than 15% of the
electricity is produced from solar PV panels. Solar panels in particular are generally
not equipped with automatic power-factor control systems, but synchronous genera-
tors are.
If relatively little power in a system is provided by large power stations, just a few
power stations need be online. Consequently, the running power stations are often
located much further from the customers than in the past. Therefore, if these few
power stations have to provide the bulk of the reactive power via overhead transmis-
sion lines, there is a high risk that the voltage to the remotely located customers will
collapse over the long lines.
Voltage collapse due to a long-distance reactive power supply is typically a local
phenomenon. It does not mean that the power plants at the generator end of the
transmission line will trip because of overloading. The long transmission lines simply
4.  Active and reactive power  95

cannot transmit the active and reactive power in case of a highly reactive load.
Consumers using the same transmission line, but at shorter distance from the power
plants, may not see the voltage drop that distant consumers experience.
The problem of voltage collapse and energy transmission restrictions can be
avoided by installing more local power capacity based on synchronous generators.
Modern moderately sized power plants offer excellent backup capacity. They can
help to provide reactive power where there is a high penetration of renewable sources
with varying output.

4.7. Conclusions
Transporting large levels of reactive power via alternating current transmission lines
over long distances leads easily to voltage collapse at the end of the line. In such cases,
the active power transport capacity is drastically reduced. This can cause a blackout.
Therefore, future electricity supply systems with a high proportion of wind turbines and
solar panels cannot only use large power plants far away from the load for load bal-
ancing and frequency control. This would have a negative impact on supply reliability
and adequacy. As a solution, smaller power stations based on modular generating units
in parallel should be positioned at relatively short distances from the load centres. This
solves the problem of local voltage collapses caused by reactive power.
5 Energy storage
It would be very convenient if electrical energy could be easily and cheaply stored
in large quantities. Storage would help to shave peaks in electricity demand and
to provide reserve capacity in the supply system. Excess electricity from wind
turbines might be used when there is no wind. Daytime power from solar panels
could be stored for the evening and abundant solar energy in the summer could
be used in the winter. Unfortunately, converting electrical energy into chemical
or mechanical energy creates significant costs and losses of energy. To support
renewable power, the most promising solution is the integration of electricity and
heat use, including heat storage.
98  Power supply challenges

5.1. The enormous challenge of energy storage


Everywhere in nature and throughout society in general, the storage of energy is essen-
tial. Plants store energy as carbohydrates. Animals store energy in fat layers for use in
times when food is scarce. Mankind learned to store crops from the fertile seasons. Har-
vesting energy from the sun, which is the basic energy provider for the world, natu-
rally occurs in batches. A typical example is the potato. Potatoes are seasonal plants,
and the potato crop normally peaks during the second half of the summer. Taking the
complete potato crop to the consumer market at once would result in very low prices
temporarily, while also creating a high risk of having no potatoes on the market long
before the new crop arrives. This is why storage is essential. A warehouse owner can buy
potatoes at harvest time while the cost is low, store them and sell at a higher price later.
Traders call the connected profit making mechanism arbitrage. Where there is competi-
tion between warehouse owners,
the price increases can be limited.
A reliable energy supply for sus-
taining the economy in modern
society has much in common with
a stable food supply for sustaining
life. In both cases, storage is indis-
pensable.
The previous chapters of this
book have shown that energy is
difficult to store as electricity.
Traditional power stations use
fuel for medium-term energy
storage. Fuel is an excellent
example of stored energy. Figure 5.1.  A reliable and affordable food supply
Hydropower-based generators require cheap and efficient means of storage.
use either the potential energy Same applies for energy supply.
stored in elevated water levels or the kinetic energy from water streams. Table 5.1.
gives some typical values for the volumetric energy density of common energy
sources for electricity production. The values given are approximations, since they
depend on the composition of the energy source. Coal, wood and oil can be stored
in bulk next to the power station. Gas is fed to the station via a pipeline or stored
locally as liquefied gas.
As an example, a coal-fired power plant with a nominal electrical output of
500 MW and a fuel efficiency of 40% consumes 500/0.4 = 1 250 MW of fuel. This
equals 1.25 GJ/s. Coal has a volumetric energy density of around 19 GJ/m3, the
exact value depending on its composition. The 500 MW power plant uses, therefore,
1.25/19 ∙ 3 600 ≈ 237 m3 of coal per hour. This is a substantial volume. Neverthe-
less, a storage area of 100 m by 100 m and a height of 10 m contains almost enough
coal for 18 days of the power plant’s full output. To illustrate this coal demand, the
5.  Energy storage  99

load capacity of a typical Panamax bulk carrier ship of Table 5.1.  Approximate volumetric energy
75000 tonnes is needed for these 18 days of power plant density of common energy sources for elec-
operation. Knowing that a 500 MW electrical output is tricity generation (G = 10 ).
9

the equivalent of the average power output from 5 million Volumetric energy
Energy source
manual labourers brings these figures into perspective. It is density
an enormous fuel stream that is needed to generate a con- Coal 19 GJ/m3
tinuous electric energy flow of 500 MW. Fortunately, fuels
Fuel oil 35 GJ/m3
such as coal, oil, and high-pressure natural gas contain a
significant amount of energy per cubic metre. Natural gas at 80 bar 3 GJ/m3
It is interesting to note that a human being needs about Liquefied natural gas 21 GJ/m3
3000 kcal of food per day in order to provide 8 hours of
Animal fat 33 GJ/m3
hard manual labour. These 3000 kcal equal 12.5 MJ, since
1 kcal = 4.1868 kJ. Three daily shifts of 5 million workers Wood 10 GJ/m3
each, altogether 15 million workers, would be needed to Water at 400 m elevation 0.004 GJ/m3
continuously rotate a generator producing 500 MW. The
total food consumption by these workers would be 225 TJ/ Uranium 235 1500000000 GJ/m3

day, or 6 800 m3 of animal fat with a calorific value of 33 Compressed air at 80 bar 0.033 GJ/ m3
GJ/m3. The power plant uses about 108 TJ per day when
it is running at full output on coal, but the workers would Hydrogen at 80 bar 0.85 GJ/ m3
need in any case some 135 TJ to sustain their bodies when
they do not work. Therefore, the fuel efficiency of the workers and that of power
plants are not so different. Nevertheless, this example illustrates the high energy
supply needed to keep a 500 MW power plant running.
For hydropower, considerably larger storage volumes are required than for fuels.
The 4 MJ/m3 of potential energy Epotential given for water at a 400 meter elevation in
table 5.1. has been derived from:
Equation 5.1.
Epotential = ρ · g · h
in which density ρ = 1000 kg/m3, g = acceleration of gravity ≈ 10 m/s2 and h = 400  m.
If the efficiency of the turbine-generator combination of a hydropower plant
equals 80%, one needs 500 MW/0.80 ∙ 4 MJ/m3 = 156 m3 of water per second
flowing from an elevation of 400 m to deliver 500 MW of electricity. By comparison,
the river Rhine has an average water flow of 2300 m3/s when it arrives at the border
between Germany and The Netherlands. The river Elbe has an average water flow of
711 m3/s when meeting the sea, while the Danube’s flow is about 1500 m3/s when
it leaves Germany. The average electricity demand in Germany in 2010 was 67 GW,
which is a factor 134 higher than the 500 MW in the example. By comparison,
average electricity demand was 473 GW in the USA and 450 GW in China in the
same year.
Looking back to the example given using coal, to run the 500 MW hydropower
plant at full output for 18 days, one needs 18 ∙ 24 ∙ 3600 ∙ 156 ≈ 245 million m3 of
water at a 400 m elevation. This equals an area of 5 km by 2 km with the water level
starting 25 m above the 400 m elevation.
100  Power supply challenges

The example shows that deriving energy from raised water levels in order to
meet the electricity consumption takes a huge amount of water. The energy density
of pipeline natural gas expressed in GJ/m 3 is roughly a factor 7 500 higher than
that of water at an elevation of 400 m. For coal, it is a factor of almost 50 000.
The idea of having 100% electricity production based on wind and solar power
with energy stored in raised water levels sounds interesting but appears difficult
to realize. Time spans without wind over large areas can last more than a week,
while solar panels provide hardly any electricity during the dark winter months.
Covering electricity demand for prolonged times without fuels remains, therefore,
a major challenge.
Nevertheless, the storage of electricity as electrical, mechanical, chemical or even
thermal energy might be economically attractive. That can be the case in deregulated
electricity markets so as to make a profit from arbitrage, and in regulated markets so
as to lower costs for consumers. The economics depend on the application. It makes
a great difference whether the storage is used for:

• Short-term applications, such as reserve capacity and frequency regulation;


• Medium-term applications, such as peak shaving with daily charging and
discharging;
• Long-term applications for bridging seasonal differences in renewable
output.

The next sections of this chapter will discuss a number of storage methods.

5.2. Basic properties of energy storage devices


Energy storage devices have a number of properties that determine their applicability and
value in electricity supply systems. Basic properties are the amount of energy (MWh)
to be stored and the speed of energy charging and discharging (MW). Figure 5.2 gives
a schematic representation of an energy storage device in an electricity supply system.
The concept is such that any electricity produced but not consumed will immediately
be sent to the storage device.
By combining a converter and a charger, electrical energy is turned into the
type of energy that can be stored. An electric motor can turn electrical energy
into mechanical energy directly by, for example, accelerating a flywheel or indi-
rectly by driving a water pump or an air compressor. An AC to DC convertor
turns an alternating current into a direct current for charging a battery or for the
electrolysis of water for hydrogen production. The power capacity of the charger is
generally determined by two things: the size of the charger and the power that the
storage device can accept. In the case of pumped hydro, the water basin itself is
not a limiting factor in the speed of filling it, but the size of the combined electric
motor and pump is. For batteries however, the charge current is limited. Charging
too quickly overheats the battery and causes a permanent loss of storage capacity.
5.  Energy storage  101

Direct delivery of electricity

Electricity Electricity
production Converter/ Converter/ consumption
charger discharger
MW MW

Energy storage
MWh/GWh/TWh

Energy storage device

Figure 5.2.  A representation of an energy storage device

Faster charging can also reduce the life of a battery. Conventional lead-acid bat-
teries used in cars require slow recharging but can sustain a large short-term
discharge current when the engine starts. In other words, the charging device can
have a different power capacity than the discharging device.
The energy storage capacity depends on the size of the storage device. Batteries
consist of chemical cells. A large number of cells can be mounted in parallel to create
more capacity. For pumped hydro, simply a larger water reservoir is needed to create
more capacity.
Energy stored as heat or chill cannot be effectively converted back to electricity.
Such energy storage is rather useful for demand-side management of electricity
supply. Electric heating coils in hot water reservoirs can be switched on during times
of high output of wind turbines and low electricity demand. Refrigerated warehouses
can be extra cooled at times when electricity prices are low. A limitation here is that
many edible products have to be stored within a narrow temperature range to pre-
serve their quality.
Energy stored in fuels shows practically no decay. Most storage devices, however,
lose energy with time. Chemical batteries lose at least 1% of their full charge per day
and flywheels some 5%. By contrast, open water reservoirs can maintain their energy.
The evaporation can be more than compensated by rainfall, although this depends
on the local climate and the season.

5.3. Applications for energy storage devices


The performance requirements for storage devices depend on their intended use. Since
storage devices can both take in and give out energy, they can in principle be used in
all balancing tasks needed for securing the supply of electricity. It is the time scale espe-
cially that determines the economic opportunity for a storage device. Such an opportu-
102  Power supply challenges

Table 5.2.  Balancing tasks in electricity supply systems.

Seconds Minutes Hourly Daily Weekly Seasonal


Summer/winter
Frequency Load Ramp rate Peak Wind output pattern
control following smoothing shaving smoothing smoothing

Voltage Predicton error Load Solar output Weekend Summer solar


support compensation following smoothing storage use in winter

Primary Secondary Demand Arbitrage


reserves reserves response

nity exists only if market rules allow remuneration for operating a storage facility. Table
5.2 summarises the balancing tasks for different time-scales. The deferral of grid expan-
sion and of peaking-power generator investments can be an additional benefit of energy
storage.
Figure 5.3 illustrates the possible tasks for storage systems within a 24-hour
time span. If a storage system is used for continuous frequency control, it is actually
permanently in operation. The amount of energy exchange is in this case relatively
small compared with the power capacity. For daily peak shaving, energy can be
accumulated during the night. Charging can take place from 11 pm to 6 am when
demand is low, and discharging can take place from 5 pm to 9 pm when demand is
high. Charging would take 7 hours and discharging 4 hours in this example. This
means that the storage device charger can have a power capacity that is a factor of
7/4 = 1.75 lower than that of the discharger.
Other opportunities for storage systems include compensating for differences
between forecasted and real demand, and between forecasted and real renewable
output. Electricity demand itself can be predicted quite well based on historic
demand patterns and weather forecasts. Large temporary deviations can, however,
occur between the predicted and actual wind power. This is especially true when
the wind speed changes rapidly from high to low or vice versa. A difference of fifteen
minutes between the forecasted wind speed and the actual wind speed can make a
huge difference in the output of a wind farm. The prediction error might be positive
or negative, so the size of the charger and the discharger has to be the same.
Frequency control might be a special application for short-term energy storage.
In market-based supply systems, frequency deviations occur especially each time a
new trading time begins, say every 15 or 60 minutes. The injection or absorption of
energy at those moments can minimise frequency deviations.
Wind park output can remain high for a number of consecutive days, and then
be followed by some 10 days of little or no wind at all. Where 20 % of the average
electricity demand in an area has to be covered by wind energy, the installed wind
turbine capacity should be 80% of the average power demand if the capacity factor
of the combined wind turbines is 25%. This can be simply calculated: with a capacity
factor of 25%, on average 0.25 ∙ 80% = 20% of electricity demand is produced by the
wind turbines. In the case of high winds covering a large area, the combined output
5.  Energy storage  103

Peak shaving
Wind prediction
error compensation Peak load
Power demand

Intermediate load Ramp rate


smoothing

+ Frequency control
Base load + Voltage control
+ Contingency reserve
+ Arbitrage

0 2 4 6 8 10 12 14 16 18 20 22 24
Time of the day (hours)

Figure 5.3.  Possible energy storage applications in time scales up to 24 hours.

of the wind turbines can provide more power than the instantaneous demand, espe-
cially during the night. Imagine that the winds are so high that 90% of the installed
wind turbine capacity is running at full output. That means that 0.90 ∙ 80% = 72% of
the average power demand is produced by wind. During the night, power demand
can easily be less than 72% of the average demand. Temporary storage of part of the
oversupply of wind power output would help to leave space for some dispatchable
power generators in the grid, and the energy stored could be used later during times
of little wind. The charging power of the storage device would be higher than the
discharge power in this case.
If the excess summertime output from solar panels would be stored for use in
the winter, the storage capacity would have to be large. Enough energy storage for
at least 3 months is needed in a case like this. Systems such as flywheels, batteries
and compressed air can immediately be discarded for such applications because
their storage capacity is limited. Also, hydrogen cannot be used for long-term storage
because of its low volumetric energy density.
Some storage systems are by nature bi-directional. This means that they can
immediately switch from charging to discharging. Lead-acid batteries have this
ability. For pumped-hydro and compressed-air storage, it takes some minutes to
switch mode. The bi-directional capability stops when a storage system is full or
empty. Identical restrictions apply for hot water and chill storage systems. If the
temperature limits for hot water reservoirs or refrigerated products are reached, such
temperature-based systems lose their storage or dispatch capability. Demand-side
management systems generally have a lower capacity during the night than during
working hours, simply because there is less electricity demand during the night.
104  Power supply challenges

5.4. Methods and costs of energy storage


5.4.1. Lead-acid batteries
Price statements for energy storage systems can be confusing. A lead-acid battery will
be used to illustrate this. Lead-acid batteries are a proven technology and require a low
investment compared with, for instance, lithium-ion batteries.
Our example is a 12 V and 80 Ah battery, typically used in cars. Based on its name-
plate, one might presume that a 12 V and 80 Ah battery has a maximum amount of 12
∙ 80 = 960 Wh energy stored, equalling almost 1 kWh. The retail price of such a bat-
tery is about 75 €. At first sight, the storage component of
the lead-acid battery-based storage system would therefore
cost about 75 €/kWh. This is however not the case, as will
be explained below.
The recommended charge current for a 12 V 80 Ah
battery is some 8 A, so an AC to DC charger of 8 A ∙ 12
V ≈ 100 W is needed. Such a charger might cost 25 €.
Full charging with 8 A takes roughly 11 hours, since the
charging efficiency is 90%. Charging the battery with too
high a current will release hydrogen and oxygen, resulting
in a loss of liquid in the battery. Too high a current leads
also to overheating and shortened battery life. Figure 5.4.  A conventional
The standard maximum discharge current of a lead- lead-acid battery.
acid battery expressed in ampere is about three times the
nameplate capacity in Ah, in this case 3 ∙ 80 = 240 A. The discharging power of the
battery is therefore 240 A ∙ 12 V = 2.7 kW. A DC to AC discharger of such a capacity
might cost 50 €. The given discharge capacity of 2.7 kW does not mean however
that the battery can use its full charge of energy of 0.96 kWh to supply this power.
If that would be the case, the 2.7 kW could be delivered during 0.96 kWh/2.7 kW
≈ 36/100 hour = 21 minutes. The nameplate capacity in Ah is, however, only valid
when discharging the battery more slowly, during a time span of 20 h. That would
mean a discharge current of 4 A instead of the maximum 240 A. However, there is
another restriction, since discharging the battery deeper than 50% of its maximum
storage capacity will drastically reduce its life. Therefore,
the practical storage capacity of a 12 volt, 80 Ah lead-acid Table 5.3.  Investment in a 12 V 80 Ah
battery in dynamic operation is less than 0.5 kWh. This lead-acid battery-based storage device.
means that the real specific capital investment in a lead- Item €
acid battery for storage use in an electricity supply system
Battery 75
turns from the initial purchase price of € 75/kWh to more
than € 150/kWh. Charger 25

This leads us to storage system costs. The battery plus Discharger 50


the charger and discharger require an investment of some
Total 150
75 € + 25 € + 50 € = 150 € for an effective 0.5 kWh storage
5.  Energy storage  105

Direct delivery of electricity

Electricity AC/DC DC/AC Electricity


production converter converter consumption
charger 100 W discharger
2.7 kW

Energy storage
1 kWh = < 0.5 kWh
effective

Car size 12 V 80 Ah lead-acid accumulator

Figure 5.5.  Example of a lead-acid energy battery for use in electricity supply systems.

(see table 5.3. and Figure 5.5). This definitively does not mean that the user has to pay
€ 150 for each kWh of electricity from the battery. If the system can sustain a max-
imum of 2 000 charging/discharging cycles, and no interest on invested capital is used,
the cost per cycle of delivering the 0.5 kWh during each cycle equals 150 €/(2 000 ∙ 0.5
kWh) = 0.15 € /kWh. This has to be added to the price of purchasing the electricity. If
the electricity for charging costs 5 €cnt/kWh, one has to add 1 €cnt to this because of
the 90% efficiency of the charging and of the discharging processes. The total price of
electricity from the storage device is then 15 + 5 + 1 = 21 €cnt/kWh, i.e. more than 4
times the purchase price of 5 €cnt/kWh. As mentioned before, this price presumes that
the capital invested in the storage system is interest free and that no maintenance and
operation costs are incurred.
Presuming a discount rate of 5% and a system life of 5 years, the annual capital
costs equal 23.1% of € 150 = 34.65 €, or about 173 € for the 5 year time span. For
2 000 full cycles in this time span, meaning again a combined energy output of
2 000 ∙ 0.5 kWh = 1 000 kWh, this renders specific investment costs of 17.3 €cnt/
kWh. That results in an output electricity price of 17.3 + 5 + 1 = 23.3 €cnt/kWh. A
lead-acid battery also needs regular inspections of the level of the liquid covering the
electrodes. That might take two man hours over the life span of the battery at a cost
of say 75 €. This adds 7 500 €cnt/1 000 kWh = 7.5 €cnt to the electricity costs. In
addition, a lead-acid battery loses charge at a rate of 10% of its full charge per month.
For a time span of 5 years, this equals 600% of its full charge, or 6 kWh. In our
example of 5 €cnt purchase costs per kWh, this decay loss amounts to only 60 €cnt
in 5 years which is close to negligible. To summarize, the application of this battery
for some 2 000 cycles with a depth of discharge of 50% will add almost 26 €cnt to
the original kWh price of 5 €cnt. After 5 years of use, batteries are hazardous waste
and their treatment may create additional costs. And, of course, after every 5 years
new batteries would have to be acquired.
106  Power supply challenges

Using the lead-acid battery for frequency control, providing rather short-term
power delivery and short-term power absorption based on a very small depth of
discharge, the technical life of the battery might be 10 years. During that time span,
some deterioration of the battery will occur due to internal corrosion. The nominal
charging power of the battery in the example was shown to be 100 W and the price
of equipment 150 €. This means that the investment in
absorbing power from the grid is 150 €/100 W or 1 500 €/ Table 5.4.  Summary of the total costs of
kW. The maximum temporary discharging power equals electricity from a lead-acid battery with 2000
2.4 kW, resulting in a discharge power investment of only cycles and an effective storage capacity of
0.5 kWh.
€150/2.4 = 62.5 €/kW. These are figures to consider when
using storage technology for short-term balancing such as Costs item €cnt/kWh
frequency control. By comparison, a gas-fuelled generator Capital costs 17.3
can do the same job and might cost about € 600/kW.
The power ramp-up and ramp-down times also Maintenance costs 7.5

play a role in estimating the value of energy storage in Charging energy costs 5
grid systems. Electro-chemical battery systems respond Efficiency loss costs 1
immediately with maximum capacity, which makes them
Total costs 30.8
very suitable for primary reserves. This is especially true
in island operation and emergency supply systems. For
short-term balancing, the size of the batteries seems less important. However, their
use is restricted to the point where at the battery has reached its minimum charge
level, or to the point where the battery is fully charged.
The key performance indicators of lead-acid-battery based energy storage systems
are summarized in Table 5.5. These figures are indications and detailed specifications
differ, depending on equipment size and supplier.
The key performance indicators (KPIs) are needed to investigate the realistic
applications of a storage technology. The KPIs given in table 5.5. reveal immediately
that storing solar-PV based energy accumulated in the summer, May – August, to
be used in the winter, November – February, is never economic with lead-acid bat-
teries. It would mean using the storage facility for only one cycle per year with an
average storage time of six months. Half of the stored energy would be lost due to
decay. With a discount rate of 5% and a maximum system life of 10 years, the costs
per kWh delivered would be at least € 45.

Table 5.5.  Example of the key performance indicators of a small lead-acid battery
based energy storage system.

Effective Maximum Charge Discharge. Operations


Turn- Energy power- Ramp-up power- Ramp Turn-
energy System depth of + around
around decay life based time based down time
storage efficiency discharge maintenance time
investment investment investment costs

€/kWh % %/month cycles % €/kW min €/kW min % of min


capital

150 80 10 2000 50 1500 0 56 0 10 0


5.  Energy storage  107

As shown above, using the battery for daily peak shaving with energy accumu-
lated during the night would add “only” about 25 €cnt to the kWh price. Frequency
control on a per-minute basis using the high discharge power might offer a more eco-
nomic opportunity, but the low charging power of lead-acid batteries is a bottleneck.
One conventional application of such batteries in electricity supply systems is to pro-
vide energy for the starting motors of engine-driven generator sets. Another common
use is in emergency applications in cases of grid failure. Batteries are needed during
the interim period when fuel-based generators are starting up.
A proven application of lead-acid battery-based storage is for off-grid use in
combination with solar panels at locations without an electricity grid. In electricity
supply systems, the best application might be for primary control reserves in the
case of power plant trips. The batteries could immediately supply their maximum
power when needed, while the storage capacity needs only to be enough for about
10 minutes until the secondary reserves start working. Another application could be
to compensate for the loss of inertia caused by many indirectly coupled renewable
energy sources in the system.

5.4.2. Lithium-ion batteries


Lithium-ion batteries are best known for their use in hybrid and fully electric cars. In
these batteries, one electrode is made of graphite while the other one is lithium-metal
oxide. Their fast charging rate when compared with lead-acid batteries and their factor
seven higher energy content per kg make them suitable for mobile applications. When
recovering kinetic energy from a vehicle during deceleration, fast charging is necessary.
Both the charging and discharging power of a Li-ion battery relative to its
energy storage capacity is about 1 W/Wh. In comparison, for a lead-acid battery the
optimum charging power is about 0.1 W/Wh and the short-term discharging power
is 2.7 W/Wh. These are average values and may differ in individual systems. Just
as in lead-acid batteries, the energy stored in Li-ion batteries decays with time. The
decay in Li-ion batteries depends very much on temperature; at 20 °C it is about 8%
per month but at 50 °C it is already 23%. Their allowed depth of discharge (DOD) is
some 80%, which is better than the 50% DOD of lead-acid batteries.
The price of Li-ion batteries is currently around 500 €/kWh. But the price of a
utility-scale storage system appears to be much higher. A UK project launched in
2013 in Bedfordshire of 10 MWh capacity intended for peak shaving, renewable
energy accommodation and grid-expansion deferral, is said to require an invest-
ment of £ 18.7 M. This equals 2 175 €/kWh of storage capacity. Apparently, the
charging and discharging equipment connecting the batteries to the high-voltage
grid, together with the control and protection devices, requires this level of invest-
ment money. As with lead-acid batteries, Li-ion batteries can accept a limited
number of deep discharging cycles. Currently, replacement is needed after some
1 500 to 6 000 cycles. Again, after this, new batteries will have to be installed and
the old ones treated as hazardous waste.
108  Power supply challenges

5.4.3. Pumped hydro


According to the World Energy Outlook 2013 publication from the International Energy
Agency, hydropower-based generation produced 3 566 TWh of electrical energy in the
year 2011. It is, therefore, by far the largest renewable energy source in the world, meeting
16% of global electricity needs. The combined installed power capacity equalled about
1 TW. The average utilisation factor of hydropower is therefore 3 566 TWh/8 760 h ∙
1 TW = 41%. One of the reasons for this relatively low utilisation factor is the variation
in demand over the course of a day and varying production over the seasons. In some
regions, no water is available during part of the year. The Three Gorges Hydro facility in
China is an example of an impressive 22.5 GW project intended to produce 100 TWh
per year, reaching a utilisation factor of 51%. Figure 5.6 shows the eight countries that
have the largest electricity production from hydropower in the world. In Norway, Ven-
ezuela, Brazil and Canada, a substantial fraction of the total electricity demand is met
by hydropower.
Hydropower facilities are concentrated in areas with high precipitation and sub-
stantial differences in elevation. In countries where much electricity is generated by
hydropower, pumped storage is generally not needed. Conventional hydropower is
excellent for balancing electricity production and demand. Pumped storage, in com-
bination with hydropower, might only help to raise the water level in basins during
the rainy season when there is excess energy from run-of-the-river generators.
Providing facilities for pumped hydro storage in countries that have to depend
primarily on fluctuating renewable sources such as solar and wind, for reducing
fossil fuel consumption is not easy. Such countries do not have much hydropower, at
least not yet, and finding space for substantially large elevated water level reservoirs

1 000 100
% of total national electricity use
Electricity from hydropower

800 80

600 60
(TWh)

400 40

200 20

0 0
a

il

ia

ay

la
az
in

ad

di
US

ss

ue
rw
Ch

In
Br

Ru
n

ez
No
Ca

n
Ve

Figure 5.6.  The eight countries with the highest electricity production from hydropower
and its fraction of total electricity demand (approximate figures, data for the year 2009).
5.  Energy storage  109

is often problematic. Currently, the globally


Upper
installed capacity for pumped hydro storage is Dam
reservoir

estimated to be between 90 GW and 127 GW,


according to diverse literature sources. This
equals just 3% of the world’s installed power
generation capacity.

e
rat
ne
In 2009, the members of Eurelectric Tunnel

Ge
penstock
(comprising almost all the European energy
companies) produced about 550  TWh from power-
house
Lower
hydropower, equalling 16% of the electricity

p
reservoir

m
Pu
demand in the area. The capacity was 198 GW. Pump- Typical pumped-
Pumped storage in the Eurelectric area is cur- turbine storage developement

rently some 35 GW, with a stored energy of


2.5 TWh. This amount of stored energy is only Figure 5.7.  A typical pumped storage facility.
sufficient to deliver the 35 GW for three days.
The averaged power demand in the Eurelectric area was about 400 GW in 2009. The
hydropower plants are primarily in Scandinavia, the Alps and the Pyrenees.
Imagine now Germany, with an electric energy demand of 590 TWh in 2010.
This converts into an average power demand of 590 TWh/8760 h ≈ 67 GW. If wind
power would cover 50% of Germany’s electricity demand in the winter and pumped
hydro could fill a gap of 10 windless days, it would require at least 67 GW ∙ 10 ∙ 24 h∙
0.5 = 8040 GWh of energy stored in pumped hydro.
With water levels at their maximum, Germany currently has 30 GWh of pumped
hydro storage. This means that Germany would need more than 250 times the cur-
rent storage volume. The connected power output capacity should be around 35 GW,
which is only a factor of 5 higher than the current output capacity of 7 GW. This
reveals that the current capacity is only intended to provide short-term power. With
7 GW of output capacity, the 30 GWh of stored energy is consumed in just 4 hours.
As a consequence, if Germany wants to smooth wind power with pumped hydro,
the logical solution is to purchase storage volume from its neighbouring countries
Norway, Sweden and Austria.
The scale issue is evident also in storing solar power with pumped hydro. In the
summer of 2013 Germany had about 40 GW of installed solar PV. With a capacity
factor of 9% over the year, the cells produced 31.5 TWh. If 10 TWh of summer solar
power would be stored for the winter, a storage capacity higher by a factor of 333
than the current capacity of 30 GWh would be needed.
Depending on the local situation, the investment for the energy storage com-
ponent of pumped hydro ranges between 30 €/kWh and 60 €/kWh. The required
pumps, electric motors, turbines and generators cost between 500 and 750 €/kW.
The energy consultancy DNV-KEMA has revealed plans for an artificial energy
storage island off the coast of The Netherlands. The power capacity would be 1.5 GW
with maximum of 20 GWh energy stored. This converts to a maximum full-load
duration of 20 GWh/1.5 GW, equalling about 13 hours. Therefore the purpose is
110  Power supply challenges

primarily daily balancing. The estimated total Current pumped hydro


storage: 30 GWh
costs of the island are between 1.3 and 1.6
billion Euros. If we use the price of 500 €/kW
mentioned above for the energy converters, the
share of the charging and dis charging equip-
ment for 1.5 GW would be 0.75 billion Euros.
Belgium is also considering building such an
Capacity of pumped
island. Again, because of the lack of storage hydro needed to
store one third of
capacity these islands will not help in backing- solar energy in
Germany
up several windless days.
The technical life of water storage basins
can exceed a century, while the charging and
discharging equipment might last 30 to 40
years. The turn-around efficiency of pumped
hydro is about 75 % for loads over 70%, and the
depth of discharge is about 90%. Annual opera-
tion and maintenance costs might amount to
3% of the invested capital. In the case that the Figure 5.8.  To store 10 TWh of the 31.5 TWh of solar
energy island’s 20 GWh storage would be fully energy produced in Germany annually, the pumped
used every day for peak shaving, the added hydro storage capacity would have to be a factor 333
costs per kWh would only be some 3.5 €cnt greater.
for a discount rate of 5%. Some evaporation
will occur depending on the water temperature and the relative humidity of the air,
but rainfall will compensate. The actual energy loss is probably negligible. Switching
between charging and discharging for pumped hydro takes around 4 minutes, while
ramping up and down the full range of some 20% to 100% capacity takes up to
1 minute.
If the storage island would be used for only a single charging and discharging cycle
per year, as in the case of moving solar energy from summer to winter, the costs per
kWh delivered would rise to about 750 €cnt.
It will be clear by now that even large
pumped-hydro storage facilities are primarily
intended for daily peak shaving and not for long-
term energy storage.

5.4.4. Flywheels
Flywheel storage utilises fast rotating elements for
accumulating energy. The rotor is accelerated, or
‘charged’, by an integrated electric motor that acts
as a generator when the system discharges. The Figure 5.9.  Image of a planned energy island in the
typical rotating speed is 20000 rpm. To reduce fric- North Sea, the so-called Plan Lievense. Courtesy of
DNV GL, Lievense CSO and Gebroeders Das.
tion losses, the rotors are positioned in a vacuum
5.  Energy storage  111

chamber and magnetic bearings are used. Typically, the amount of energy stored is 25
kWh with a discharge capacity of 100 kW. The typical dimensions of such a flywheel
would have a height of 2 m and a diameter of 1.2 m. The ramp up time to full capacity
is about 1 second.
Flywheels as storage systems are still under development. Their properties could
be suitable for frequency regulation and spinning reserve. Unlike batteries, flywheels
do not have a limited number of charging cycles. In that respect they might be a pre-
ferred option for frequency regulation. However, their current size is too small for
large-scale electricity supply applications.
Most probably, flywheel-based energy
Upper Axial
storage will be limited to smaller, niche- vacuum electromagnet
chambers
type applications.
Lower Upper radial
vacuum electromagnet
chambers
5.4.5. Compressed air
Carbon fiber Patented
composite molecular
flywheel vacuum
Pressurised air can release its potential sleeve
energy by expanding over a turbine or Motor- Synchronous
generator reluctance
reciprocating expander. Pressurised air can stator 4 pole m-g
rotor
also supply a combustion turbine with air 2” thick
steel Lower radial
for the combustion chamber. The energy housing electromagnet
density of compressed air at a pressure
of 70 bar (7 MPa) is about 29 MJ/m3,
which is quite low compared to most other Figure 5.10.  Illustration of a flywheel energy storage
means of energy storage as shown in table device. Image source: Powerthru.
5.1. For storing an amount of energy that
might be useful for electricity supply systems, large underground caverns or aquifers
are required. One problem with compressing ambient air is that unless the air is cooled
during compression, it can reach a temperature of 700 °C when being compressed to
70 bar. This drastically increases the power demand for compression. If the air cools off
during storage, the pressure drops in proportion to the absolute temperature. Cooling
from 700 °C to 100 °C reduces the pressure by a factor 2.6. This is why practical instal-
lations use intercoolers during the compression process. The heat from the intercoolers
is then released to the atmosphere. With so-called adiabatic storage systems, the heat
released during the compression process is also stored and is used for heating the air
to create more air volume before it expands. The storage volume has preferably to be
filled against a fixed pressure created by a water column, otherwise the compression and
expansion equipment has to operate across a wide pressure range.
Figure 5.11 is an illustration of a compressed-air energy storage system (CAES).
There are two compressed-air energy storage installations (CAES) is the world: in
Huntdorf, Germany and McIntosh, Alabama, USA. The 110 MW output McIntosh
plant, built in 1991, requires 0.69 kWh of electricity for compression and 1.17 kWh
of fuel energy to produce 1 kWh of electrical output. The energetic efficiency of the
plant therefore equals 54%. However, proponents of the technology state that the
112  Power supply challenges

Electric
grid

Compressor Expansion
+ cooler turbine

Air
intake

Clutch A Electric Clutch B


Closed during motor/ Closed during
compression generator expansion Fuel
input
Heater

Heat
exchanger
Valve A Valve B
Open during Open during Exhaust
compression expansion

Compressed air 70 bar

Water at 70 bar

Figure 5.11.  Diagram of a CAES energy storage system.

energy stored in fuel can never be fully converted into electric energy. They use by
convention a standard 50% efficiency for turning fuel energy into electricity, thus
ending up with an alternatively defined turn-around efficiency of 78.5%. The 1978
built, 290 MW output Huntdorf plant has a turn-around efficiency of only 62.5%,
even with this positive definition.
In the example of a CAES the size of the McIntosh plant, providing 110 MW
output requires a fuel supply of 117 MW and a compressed-air supply of 69 MW.
If the air is supplied at a pressure of 70 bar and a consequent energy air density of
29 MJ/m3, the connected air flow equals roughly 69/29 = 2.4 m3 per second. To
run the plant during 4 hours for peaking applications requires a stored volume of 4 ∙
3600 ∙ 2.4 ≈ 34560 m3. This is a storage volume of about 33 m ∙ 33 m ∙ 33 m. It will
be clear that only underground caverns and geological structures are large enough
to hold the required volumes. Apparently, using CAES for long-term storage of large
quantities of energy is not realistic. The main application might be found for use in
time spans less than one day for peak shaving.
Compressed-air energy storage in the discharging mode has start-up and
ramping-up times equal to that of an open-cycle gas turbine. This would allow it to
be used for secondary reserve control, but not for primary reserves as in the case of
batteries. Also, its output load range is comparable with that of a gas turbine, with
a rapidly decreasing efficiency below 70% load. When starting the charging mode,
the torque on the driving motors has to be gradually increased with a compressor
bypass. Its power absorption capacity fully depends on the size of the compressor.
If a cavity of 34560 m 3 has to be filled during 8 hours to store 34540 m 3 ∙ 29 MJ/
5.  Energy storage  113

m 3 = 1001660 MJ ≈ 1 TJ, the energy flow into the cavity equals 1001660 MJ/ (8
∙ 3600) = 34.8 MW. For a presumed isentropic 85% efficiency of the compression
process and a motor efficiency of 96%, the power consumption during charging
will be 42.5 MW.
The costs of such a CAES system depend heavily on the availability of sufficient
storage volume. The compression equipment might costs some 500 €/kW, the tur-
bine, heat exchanger and generator some 1000 €/kW, and the high-pressure valves
and control equipment some 10 M€. These costs are estimates for a well-established
product; test sites with initially unique equipment might be factors more expensive.
The above mentioned costs add up to a total of almost 150 M€, or 1360 € per kW of
output power. To that has to be added the development work for the storage volume,
which depends to a large extent on the availability of a suitable site.
If the system is used during 1000 hours a year and the technical life is estimated
at 30 years, the annual capital costs (Fixed Charge Rate FCR) of 150 M€ investment
for a discount rate of 5% are 0.065 ∙ 150 M€ = 9.75 M€, since here:

(1+0.05)30
FCR = 0.05 · = 0.065 Equation 5.2.
(1+0.05)30–1

The discount rate is the interest rate to be paid for the invested capital.
Table 5.6 summarises the costs of the 110 MW CAES plant example operating
for 1000 full output hours per year. The costs per kWh of delivered electricity are
about 15 €cnt/kWh. It should be stressed that no investment and operation costs for
the storage space are included here. Also, any emission charges are excluded. If the
electricity purchase price during charging would be a typical 6 €cnt/kWh instead of
the very cheap 3 €cnt/kWh, this would add another 2.1 €cnt to the costs per kWh.
In comparison, a peaking plant running on natural gas can produce electricity for
less than 10 €cnt/kWh for the same boundary conditions as in table 5.6.
When we compare the costs for daily peak shaving of a CAES with that of
pumped hydro, CAES can be more expensive by a factor of 5 to 20.

Table 5.6.  Example of the estimated electricity output costs of a 110 MW CAES in the
case of 1000 full load running hours per year and an electricity purchase price of 3 €cnt/kWh

Costs item Condition Costs

Capital costs 5% discount, 30 years life 8.9 €cnt/kWh

Fuel costs 6 €/GJ fuel, 1.17 kWh fuel/kWh 2.5 €cnt/kWh

Input electricity 3 € cnt/kWh input, 0.69 kWh/kWh 2.1 €cnt/kWh

Operation/maintenance 1.5 €cnt/kWh

Total costs output electricity (excluding storage volume costs) 15 € cnt/kWh


114  Power supply challenges

5.4.6. Power to gas


The natural gas sector in Europe recently ventilated ideas on using the large pipelines cur-
rently used for transporting natural gas, for storing excess electrical energy as hydrogen.
Hydrogen can be produced by electrolysis of water. Electrolysis is barely used for bulk
hydrogen production in industry because of the poor energetic efficiency and the high
costs of purchasing electricity. Hydrocarbons, primarily natural gas and oil, are used
for producing hydrogen for fertiliser factories and refineries. The theoretical maximum
efficiency to be reached by the current technologies in producing hydrogen with elec-
trolysis is only about 70 %, despite extensive research carried out this far.
If hydrogen were to be injected into high-pressure gas transmission pipes, at typ-
ical pressures of 6 to 8 MPa, the process of compression would require a substantial
amount of energy. Hydrogen is always said to have a high specific energy content,
but that refers to its mass based energy density of 120 MJ/kg. The volumetric energy
content of hydrogen at 0 °C and 101.325 kPa pressure is only 10.78 MJ/m3, which
is on average only about a quarter of that of natural gas. The point now is that com-
pressors are volume-based machines. Compressing 1 kg of hydrogen from its ambient
pressure to 6 MPa for pipeline injection takes about 7 MJ of compression energy in
a multi-stage process with intercoolers. The electric motor driving the compressor
therefore consumes an amount of electricity equal to 6 % of the energy stored in the
compressed gas. That further reduces the efficiency of the process.
Also, transporting hydrogen by pipeline consumes more energy than trans-
porting natural gas. The reason is again the low volumetric energy density of
hydrogen, meaning that close to four times more hydrogen gas has to be trans-
ported for the same amount of energy than in the case of natural gas. Clearly, the
transport and storage capacities of natural-gas pipeline systems are drastically
reduced if hydrogen would make up a large fraction of the gas mixture. If the
intention is to convert the stored hydrogen back to electricity, the theoretically
maximum conversion efficiency is 60 %. The theoretical maximum turn-around
efficiency would be 70 ∙ (1- 0.06) ∙ 0.60 = 0.04 = 40 %. In practice, however, this
efficiency might be as low as 25 %.
Should hydrogen be produced only in times of peak electricity from renewable
sources with associated low electricity prices, the utilisation factor of the required
equipment might be as low as 4%. Investing in equipment with such a low utilisation
can hardly be profitable. If the produced hydrogen were to be injected into natural
gas streams, the resulting blend should not show large variations in composition.
Almost all gas applications suffer from variable gas composition, because it becomes
more difficult to optimise the combustion process with respect to fuel efficiency and
emissions. At the same time, gas meters measure volume streams and, therefore, a
misreading of the amount of delivered energy will occur if the volumetric energy
content varies.
Nevertheless, the world’s economy needs much hydrogen as feedstock for the
chemical industry. According to marketsandmarkets.com, 53 million tonnes of
5.  Energy storage  115

hydrogen were produced globally in


45

Ccaloricif value (MJ/m3)


2010. The energy chemically stored in Upper calorific value
40
that amount of hydrogen is 6.35 PJ (or
6.35 million GJ), or 151 million tonnes 35
Lower calorific value
of oil equivalent (toe). This is roughly 30
1.2 % of the global energy supply in 25
2010, and represents a great deal of 20
energy. About half of the hydrogen is 0 10 20 30 40 50
Volumetric percentage of hydrogen in a
used for making ammonia for fertilisers, reference natural gas (%)
while the other half is used mainly in
refineries to build lighter components Figure 5.12.  Reduction in the calorific value of natural gas
from crude oils. Most hydrogen is through blending with hydrogen.
derived from natural gas, oil, and coal
via steam reforming. The equations that govern the process are:
HC + H2O –> CO + 2H2
Equation 5.3.
CO + H2O –> CO2 + H2
Steam reforming has an energy-efficiency of 65–75 %. As mentioned earlier,
electrolysis is barely used for bulk hydrogen production because of the poor full-
cycle energy efficiency of fuel to electricity and electricity to hydrogen. However, the
chemical industry could turn their hydrogen production into a hybrid process with an
option to switch from natural gas to electricity, thereby profiting from very low elec-
tricity prices during times of excess output from renewable electricity sources. Such an
approach could also reduce the chemical industry’s greenhouse gas emissions.
In summary, intermixing variable amounts of hydrogen with natural gas for
energy storage would have very low energy efficiency and would deteriorate the gas
quality. Methanisation of hydrogen would produce a higher quality gas, but such a
process has even lower energy efficiency than the production of hydrogen. A better
application for hydrogen produced with excess electricity might be found in the
chemical industry.

5.4.7. Heat and chill


Converting electrical energy into heat or chill for energy storage reduces thermody-
namically the exergetic value of the energy. Electrical energy has an exergetic value
factor of 1, since theoretically it can be converted back into mechanical energy with
100% efficiency. Even though producing heat with electricity reduces the exergetic
value of electricity, it is often worth it. The need for low temperature heat in the world
is large, for industrial processes as well as for the heating of buildings, for cooking and
for sanitary water. Heating water with natural gas also means using an energy source
of high exergetic value. Furthermore, if an electric heat pump is used, the amount of
heat energy produced can be a factor three to six higher than the energy input from
electricity.
116  Power supply challenges

Heat and chill can be excellently stored in water at relatively low cost.
Examples exist of heating the contents of underground water reservoirs with heat
from solar collectors in the summer for use in the winter. A district heating system
in Austria owned by EVN uses a 50000 m3 tank for heat storage to balance the
heat output from a large combined heat and power plant with heat demand. Since
the specific heat capacity of water equals 4.185 kJ/(kg K), cooling the contents of
the tank from 94 °C to 60 °C releases 50 000 000 kg ∙ (94 – 60)K ∙ 4.185 kJ/(kg
K) = 7114500000 kJ ≈ 7.1 TJ. Heating the contents again with excess electricity
requires about 2 GWh, since 1 kWh equals 3.6 MJ so that 2 GWh is 7.2 TJ. Dissi-
pating 2 GWh during a time span of 4 hours means that the tank can absorb 2000
MWh/4 h = 500 MW of electricity during 4 hours while heating the contents
again from 60 °C to 94 °C. The capital investment for such a tank is about 100 €/
m 3, or about 3 € per kWh storage capacity.
Many local combined heat and power (CHP) installations of 2 to 20 MW of elec-
trical output are equipped with heat storage tanks. Applications are primarily found
in district heating systems and greenhouses. Modern home heating systems also
use heat storage, sometimes in combination with solar heat collectors. In Denmark,
heat storage tanks are increasingly being equipped with electrical heating coils for
accepting cheap excess electricity from wind turbines. Heat storage is an excellent
way of smoothing the variable output from renewable energy sources. This reduces
both fossil fuel consumption and greenhouse gas emissions.
Heat pumps help improve the electricity to heat conversion effectiveness when
low output temperatures are required. Underfloor heating in combination with heat
pumps can yield a so called coefficient of performance (COP) that easily exceeds

Figure 5.13.  A 50000 m3 heat storage tank owned by EVN in Theiss, Austria, with a
capacity of up to 7.2 TJ.
5.  Energy storage  117

three. A COP of three means that each kWh of


electricity supplied to the system results in three
kWh (10.8 MJ) of heating. This is a higher by
a factor of three than direct electrical heating.
Electrical heat pumps are therefore considered as
the preferred providers of heating in modern well-
insulated buildings.
A huge number of warehouses in the world use
chilling for preserving the quality of perishable
products. Melting ice of 0 °C into water of the
same temperature consumes 334 kJ/kg of heat,
equal to the amount of energy needed for heating
water from 10 °C to 90 °C. This is the so-called
latent heat of melting. Therefore, freezing an
amount of water with excess electricity is again
an excellent way of storing energy. Turning a
liquid into a solid, and vice versa, is called phase Figure 5.14.  Melting ice
changing. consumes heat.
One example of phase changing is the cooling
power of a popsicle, a low-calorie alternative to ice cream. The specific sensible heat
of ice is 2.1 kJ/ (kg K). If we presume a mass of 100 g and an initial temperature
of –10 °C, the consumed ice takes 10 ∙ 0.1 ∙ 2.1 = 2.1 kJ of heat from the human
body to warm up to 0 °C. Melting the popsicle takes 0.1 ∙ 334 = 33.4 kJ. Heating
the consumed ice to a body temperature of 37 °C takes 37 ∙ 0.1 ∙ 4.2 = 15.5 kJ. The
total heat input from the body to the popsicle is, therefore, 2.1 + 33.4 + 15.5 = 51 kJ.
This equals a heat consumption of 12.2 kcal, or only 0.6% of an average daily energy
intake with food. Although melting ice takes relatively much energy, slimming by
eating popsicles is apparently not very effective. If the popsicle would contain 5 gram
of sugar with an energy content of 84 kJ, the slimming effect is even negative. A daily
intake of 17 kg of pure ice at –10 °C would be better. This amount is about 25% of
the average mass of a human body and consumes 100 % of the daily food intake for
the heating.

5.7. Discussion on energy storage


All energy storage systems discussed here are suitable only for short and medium
term storage. That also applies for storage options not dealt with in this chapter, such
as sodium-sulphur batteries, mass-based gravity systems and ocean bottom balloons.
Demand response systems with smart meters and smart appliances are also solutions
for short-term balancing only. Economically and practically, storing energy in batteries,
compressed air, flywheels, hydrogen and even pumped hydro is not realistic for time
spans exceeding a week. In reality, there are prolonged time spans where renewable
energy sources have a limited output, even over large geographical areas. Latitudes
118  Power supply challenges

above 45° have hardly any sunshine in the


winter season.

Specific co2 emission (g/MJ)


100
The large-scale application of renew-
able energy sources can drastically reduce 80

the use of fossil fuels. However, with the 60


current state-of-the-art energy storage
40
technology, a full withdrawal from tradi-
tional fuels is economically not possible. 20

The associated costs would be excessive. 0


However, even without energy storage, a

ne

il

ite al
lg e

oa
as

lo
an
ra g

gn o
)
ha

tu era

kc

(li n c
ue
op
substantial decrease in the use of fossil

et

ac
na Av

ow
yf
Pr
M

Bl
av

Br
fuels can be achieved. Backup can be

He
provided with the traditional fuels: gas,
oil, coal and nuclear fuel, which offer easy Figure 5.15.  Emissions of CO2 per fuel type, based on the
long-term storage. In particular, gaseous lower calorific value.
fuels seem to offer good possibilities,
since much gas is available as natural gas, shale gas, coal-bed methane and biogas.
The CO2 emission of methane, the main constituent of gaseous fuels, is 54.8 g/MJ,
which is low compared to coal and oil (see Figure 5.15).
The Bergermeer facility under construction near Bergen in The Netherlands
will be an example of a typical gas storage site. The gas field can store 8.4 billion
normal cubic metres of gas, of which 4.1 billion m3 is its working volume. The rest

10 000 Natural gas cavern


Storage as heat
Power investment (Euro/kW)

Compressed air
1 000 Pumped hydro
Lead-acid battery
NA-S battery
Flywheel
100 Li-ion battery

10

1
0.01 0.1 1 10 100 1 000
Energy storage investment (Euro/kWh)

Figure 5.16.  General impression of investment levels for different storage techniques.
The costs per kWh of electric energy delivered depend on the life of the storage method
and its utilisation factor.
5.  Energy storage  119

of the volume is made up of cushion


gas. For a lower calorific value of 36
MJ/m3, the working energy storage
capacity equals 4.1 · 109 · 36 =
147600000000 MJ = 41 TWh of fuel.
For an injection time of 110 days, the
average charging fuel flow equals
16.3 GW. The equipment for filling
the field restricts the injection flow
to 19.75 GW. For a withdrawal time
of 90 days, the average discharge fuel
flow equals 19.9 GW. The technically
maximum discharge flow is 26.8 GW.
If the maximum discharge fuel flow
of 26.8 GW is converted into elec- Figure 5.17.  Artist's impression of the Bergermeer gas storage
tricity with 48% efficiency, it renders facility. The site is under construction in the Netherlands.
an electric power of 12.9 GW. That is
slightly more than half the maximum output of the large Three Gorges hydro-power
plant in China. The estimated total costs of the facility are some 800 M€.
If a practical efficiency of flexible systems for converting gas into electricity of
48% is presumed, the investment for the Bergermeer facility will be 0.04 € per kWh
storage volume for electricity. For a discount rate of 5% and a technical life of 30
years, this renders annual additional costs of only 0.25 €cnt/kWh delivered, even
if the facility would be charged and discharged only once per year. Such low costs
challenge every short-term storage alternative.

5.6. Conclusions
For most energy storage technologies, frequency regulation and peak shaving seem to be
the best applications. But storing energy for more than a few days can easily more than
double electricity costs. Combining renewable energy sources with heat production in
CHP systems could offer an effective means for using excessive renewable energy. This
would reduce fossil fuel consumption substantially. Full abstinence from fossil fuels
appears to be utterly uneconomic with current state-of-the-art storage technology. Nat-
ural gas seems to offer low-cost solutions for storing energy and balancing electricity
supply and demand with a relatively low burden on the environment.
6 Costs of
producing electricity
Knowing the production costs of electrical energy is crucial for power producers
operating in competitive markets. Many cost items cannot be controlled by the
owner of a power plant. Therefore, profitability of an investment is very difficult to
predict. This is challenging for electricity producers. The cost figures given in this
chapter are examples that can be modified by the reader for particular situations.
The intention is to gain an insight into the underlying mechanisms that determine
the kWh costs.
122  Power supply challenges

6.1. Challenges in determining kWh costs


Low electricity costs are crucial for the economy. This is especially true when a country
has to import the generating equipment and the required primary energy. Power plants,
as well as electricity transmission and distribution systems, are costly long-term capital
investments. The technical life of these systems can exceed forty years and the payback
times are long.
Electricity supply systems have to comply with a large number of technical
rules and regulations that change continuously. This adds costs to the already high
basic investments in equipment. Therefore, even in a competitive electricity market,
owners of electricity supply systems need a certain amount of confidence that their
investment will pay off and ultimately provide them with a decent profit.
The short-term interference with market conditions by policy makers further
complicates the matter. One might even wonder if open markets are really the best
solution for long-term investments that have such a crucial impact upon society.
Free markets are obviously the best solutions for simple commodities that can easily
be shipped and stored, but electricity is not a simple commodity. To cope with the
uncertainty, a new keyword is emerging within the energy sector: flexibility.

6.2. Varying conditions for generating electricity


There is no universal solution available for providing reliable and affordable electricity in
a sustainable way because boundary conditions differ completely from one location to
another, and they change often. Some countries can go for easy solutions. For example,
Norway has such vast resources of hydro-power that the demand for electricity can
easily be covered in all seasons. France and Switzerland rely on nuclear power plants that
produce electricity at low marginal costs per kWh. Yet, the Fukushima nuclear disaster
has shown that the ultimate life-cycle costs of nuclear power plants can be excessive, and
the effects on the environment negative.
Power supply based on solar radiation can be effectively used in areas with much
sunshine during all seasons, provided short-term storage is available to cover the
nightly demand. As seen in chapter 5, long-term storage of electricity is prohibitively
expensive. Therefore, in most cases it is not feasible to obtain a stable and low-cost
electricity supply from renewables without using fossil fuels as backup. In particular,
for areas with substantial seasonal differences in weather conditions, fossil fuels seem
to be the only energy storage method that is practically and economically available.
It is not likely that affordable long-term-storage systems will be developed soon. Yet,
some policy makers have decided that the use of fossil fuels for electricity produc-
tion must be totally curtailed within the coming few decades. The costs and conse-
quences of such a target may not always be understood.
Fuel prices differ substantially depending upon location. In North America, an
abundant availability of natural gas resulting from the production of shale gas has
lowered gas prices since 2012. In Asia, many countries depend on gas imports and
6.  Costs of producing electricity  123

are, therefore, captive customers that


face high gas prices at levels almost equal ENDEX TFF Gas Future Reference Price, ICE Brent Index (BINDEX) Daily
Price & Henry Hub Natural Gas Future Settlement Price
to oil per MJ. Although indigenous gas
30
resources in Europe cannot fully cover Brent oil
the demand, their presence keeps gas 25

Fuel price (US$/GJ)


prices lower than in Asia. If there were 20

no issues with emissions, coal might 15


APX-ENDEX gas
maintain its role as the major provider of
10
cheap energy for electricity production.
5
Coal resources are widespread and abun- Henry Hub gas
dant in the world. 0
0 52 104 156 208 260 312 364 416 468 510
Furthermore, market conditions for
electricity differ considerable from country Weeks from January 1, 2004 till April 1, 2014

to country. Vertically integrated electricity


supply companies with a monopoly are less Figure 6.1.  The development of oil and natural gas prices in
vulnerable with respect to economic sur- the world.
vival than companies that have to compete
in open markets. The monopoly companies are often owned by the state, a province or
a municipality. Their goal is to serve the community with affordable and reliable elec-
tricity, even in remote areas. Thus, their customers indirectly own the company. Man-
agement in such companies has to ensure that the electricity supply meets the rules
set by the policy makers. All costs are recovered by charging customers with a fixed
connection fee, a capacity fee related to the maximum electric power of the connec-
tion, and an energy fee for each kWh consumed. Also private energy companies can be
vertically integrated and monopolistic. In that case, a government appointed regulator
might be employed to oversee their processes and use, for instance, benchmarking to
establish low consumer prices or to ensure maximum tax revenues.
In fully competitive markets, balancing electricity demand and production in the
supply system is controlled by a transmission system operator (TSO) who requests
a certain amount of production within a certain time span based on demand pre-
diction models. Such models use historical demand patterns in combination with
weather forecasts. Where there are many wind turbines and solar PV panels in
the system, the TSO also uses weather predictions for forecasting the renewable
electricity production. Independent power producers can offer their output to an
energy exchange, both on a day-ahead basis as well as in hourly intervals, or even
five-minute intervals.
In some market systems, electricity producers are compensated not only for pro-
viding electric energy, but also for the so-called ancillary services that are needed to
keep the system stable. These ancillary services are frequency control, contingency
reserves in case of plant failures, load following, fast ramping up and down, reactive
power supply, and forecasting error compensation. Ultimately, free markets for the
production, transportation, and retailing of electricity can only survive if the system
is sufficiently profitable for the investors.
124  Power supply challenges

Vertically integrated
electricity supply Unbundled, free market electricity supply Electricity
retail Customers
Generators company

Transmission Independent Transmission Distribution Electricity


electricity system system retail Customers
generators operator operator company
Distribution
Electricity
Retail retail Customers
company

Customers

Figure 6.2.  Two extremes: a fully integrated electricity supply system and a completely
free market electricity supply system.

Volatility in boundary conditions resulting from unpredictable fuel prices,


changing environmental restrictions, non-dispatchable electricity sources, vari-
able subsidy schemes, and frequent interference from policy makers has completely
changed the electricity supply sector. Most power industry leaders are pleading for
more stability in the rules and regulations; otherwise it becomes unattractive to
invest in power supply systems. Yet, open markets have created a substantial lobby
circuit that market players use to influence
Government Competition
politicians and bureaucrats. Opinions actions Public
opinion
sometimes have more impact on decision
TSO rules Emission
makers than technical and scientific argu- legislation
Electricity
mentation. Taxes generator Government
supported
competition
Economy
6.3. Cost analysis for different dependent
demand Fuel costs
generating techniques Shareholder Market
profits rules

Electricity generators, transmission lines,


and distribution systems cost money and Figure 6.3.  The many boundary condi-
are, therefore, capital investments. Monop- tions of an electricity generator in a free
olistic public utilities can collect fees from market.
customers to build up capital resources so that no money has to be borrowed from a
bank for investments. As a consequence, the money paid in advance cannot be used by
the customers for private investments or bank account savings. This customer money
will render lower capital costs than when the power supplier borrows capital and pays
shareholders a dividend. Additional costs are imposed by the maintenance of equip-
ment and the salaries of operators. Fuel-based generators are burdened with fuel costs,
the extent of which depends upon their efficiency and the price of the fuel. Reserve
capacity is needed for contingency reserves and for guaranteeing proper balancing in
case of demand forecasting errors and the output from renewable resources.
The ultimate production costs per kWh can differ greatly depending on circum-
stances. If a power supplier uses an existing power plant, the investment for which
6.  Costs of producing electricity  125

has been fully depreciated, it can produce


electricity at almost marginal costs. Maintenance Demolition Capital
costs costs costs
A new-built plant burdened with bank
loans and shareholder investments can
have difficulty in competing with an old
existing plant, although its fuel efficiency Fuel costs Electricity
POWER
might be much higher. This seems unfair PLANT
but it is the consequence of free markets.

6.3.1. Capital costs Operation Project Emission


costs lead costs
The capital costs per kWh produced are costs
determined by three factors: the payment
rate for capital, the life of the equipment, Figure 6.4.  The various costs related to electricity production.
and the utilisation factor of the equipment.
The payment rate for capital is by definition called the discount rate R:
R = share holder capital fraction · profit rate + bank loan fraction · interest rate
Equation 6.1.

As an example, if the shareholder fraction of the capital investment equals 25%


and a profit rate of 10% is required, while the bank provides 75% of the capital at an
interest rate of 6%, the resulting discount rate R equals:
R = 0.25 · 0.10 + 0.75 · 0.06 = 0.07 Equation 6.2.

For an equipment life expectancy of n years, the so-called fixed charge rate FCR
equals: (1 + R)n
FCR = R · Equation 6.3.
(1 + R)n –1
The fixed charge rate is then 0.075 for a discount rate of 0.07 and an equipment
life of 40 years, while for a life of just 20 years, the FCR will be 0.094. The equations
allow the readers to determine the FCR for their own possible application.
Table 6.1. gives examples of typical capital investments in different power plant
techniques. The equipment investment price can differ from country to country,
depending upon, for example, equipment costs, local labour costs, and infrastruc-
tural requirements. The additional costs caused by the lead time of each project have
been estimated by using a fixed charge rate based on a discount rate of 0.07 and an
average investment of half the equipment and installation costs during the lead time
until commissioning. Technically, the lifetime of generating equipment can be very
long, but spare parts might become obsolete and cumulative changes in regulations
might prove the technique to be inadequate in the long run.
The capital costs per kWh produced for the different generating methods as
mentioned in table 6.1., depend to a large extent on the utilisation factor of the gen-
126  Power supply challenges

Table 6.1.  Examples of typical capital investments for different generating methods.

Generating Typical equipment Project lead Investment Equipment


set type + installation time including lead life
investment time costs
€/kW months €/kW years

Hard coal 1500 40 1687 40

Nuclear 3000 60 3562 40

Gas turbine
750 24 806 40
comb. cycle
Gas engine
500 12 519 40
simple cycle
Wind onshore 1500 6 1528 20

Wind offshore 3000 12 3112 20

Solar PV cell 1700 3 1716 20

Hydro 1000 50 1175 60

erators. A utilisation factor of 100% means continuously running at 100% capacity


throughout the life of the equipment. This is physically impossible, since every tech-
nique needs maintenance, while occasionally failures occur resulting in additional
downtime. At the same time, the constant need for balancing electricity demand
and production means that not all generators can always run at maximum output.
Sometimes they run on part-load and sometimes they are even switched off because
of lack of demand.
Non-dispatchable generators based on wind and solar radiation, for instance,
are not able to produce their maximum output all the time. Wind turbines and
solar panels often have a legislation-based priority for feeding their electricity

Figure 6.5.  Every machine needs regular maintenance resulting in downtime.


6.  Costs of producing electricity  127

to the grid. However, due to the variability of wind speed and solar irradiation
their maximum output is available only during a small fraction of the time. This
is expressed in their capacity factor, which shows how many kWh they produce
in reality compared with the kWh they would produce by running at full output.
Chapter 3 shows that the capacity factor of solar PV panels can be below 10% in
countries with a relatively dark winter season. In very sunny countries, it might
reach 30%. Onshore wind turbines have a capacity factor from 15 % to 35 %.
Offshore wind turbines might have a capacity factor ranging between 20 to 45
%. However, the highest percentage applies only for optimum locations. With an
unrestricted feed-in possibility, the capacity factor immediately translates into the
utilisation factor. It is interesting to note from table 6.1. that onshore wind tur-
bines currently require almost the same capital investment per kW as solar panels.
In countries with a moderate amount of sunshine, such as Germany, the electricity
production in kWh per installed kW is about twice as high for onshore wind tur-
bines than for solar panels. This means that the capital costs per kWh produced are
twice as high for the solar panels as for the wind turbines, under the assumption
that their lifecycles are of equal length.
The capital costs expressed in €cnt/kWh for the different generating techniques
given in table 6.1. are shown in Figure 6.6 as a function of the utilisation factor. It
should be kept in mind that the costs shown are based on the investment data in
table 6.1. for a discount rate of 7%. A fully depreciated power plant that remains a
remnant of a monopolistic utility might theoretically have zero capital costs. Private
investors in solar PV panels might be happy with a discount rate of 3%, especially in
times when interest rates on savings are low and substantial subsidies for installing
panels are available. This can result in artificially low capital costs for renewable
energy sources.
It seems fair to compare the different generating techniques for the same finan-
cial boundary conditions, because subsidies ultimately originate from taxpayer
money. It is also important to know that every generating technique needs backup
power, the amount of which depends upon its reliability and maintenance needs.
Fuel-based and hydro-electric power plants can have an availability of at least 95%
and therefore require backup capacity. They can share this reserve power resulting
in 5 % additional capital costs per power plant. Solar panels and wind turbines need
close to 100% backup capacity. That easily adds 1 to 2 €cnt/kWh to the capital costs
of the mentioned renewable energy sources, even when using relatively cheap gas
engines or gas turbines for backup.
The lines in Figure 6.6 for the capital costs of generators based on wind and solar
radiation do not cover the same utilisation factor range as those of other generating
techniques. The reason is that the maximum capacity factor for solar panels is 30%,
for onshore wind 35%, and for offshore wind 45 %. In most cases, the actual capacity
factors will be much lower than these maximum values.
Generators based on wind, solar radiation, and nuclear energy have on the one
hand the highest specific capital costs, but on the other hand they need no fuel or
128  Power supply challenges

20

18
Offshore wind
Specifiic capital factor (€cts/kWh)

16 Nuclear
Solar PV
14 Onshore wind
Hard coal
12
Hydro
10 Gas turbine c.c.
Gas engine s.c.
8

0
0 10 20 30 40 50 60 70 80 90 100
Utilisation factor (%)

Figure 6.6.  Capital costs per kWh depending on the utilisation factor, based on data from
table 6.1 and a discount rate of 7%.

the fuel cost is very low. The high investment costs of nuclear and coal-based power
plants mean that the plants should preferably run at full output as much as possible.
In addition, the steam-based techniques used for these power plants are not very
suitable for frequent starts and stops and for rapid changes in output. Steam boilers
and steam turbines suffer less wear if they operate under constant physical condi-
tions. Gas engine and gas turbine-based techniques are the best solutions for peaking
power and intermediate power, since their specific capital costs stay relatively low,
even in the case of a low utilisation factor.

6.3.2. Primary energy costs


Fuel-based power plants require primary energy for producing electricity. The price of
fuel depends on production costs, profit rates, transportation fees, storage costs and
subsidies, as well as taxes. Table 6.2. gives some indicative prices per unit of primary
energy as valid in North America in 2013. Figure 6.1 showed already that large differ-
ences in the price of fuel can occur from country to country. The price of Australian
thermal coal was about 0.3 €cnt/MJ in 2013. Five years earlier, the price was about three
times higher. When cheap shale gas reached the North American markets, it became
cheaper to use gas for electricity production than coal. Gas has also less emission issues.
The decline in coal use in North America resulted in low international coal prices and,
consequently, imported coal is currently the cheapest primary energy source for power
plants in Europe. In 2013, many brand new highly efficient gas-fuelled power plants
6.  Costs of producing electricity  129

could not compete because of the low Table 6.2.  Indicative global prices of pri-
price of coal. At the same time, many mary energy in North America in 2013.
mothballed coal-fired plants were put
Fuel type Fuel price (€cnt/MJ)
back into operation.
The efficiency of converting Thermal coal 0.15
fuel into electrical energy is also a Nuclear fuel 0.10
determining factor for the fuel costs
Natural gas 0.24
expressed in €cnt/kWh. Table 6.3.
gives an indication of the fuel efficien- Light fuel oil 1.20

cies of different electricity generating


techniques running at full output. These are power plant efficiencies and not the
efficiencies related to delivering the energy to customers. Transmission and distribu-
tionlines and transformers have their losses, which means the ultimate fuel efficiency
of supplying customers with electricity from a distant large power plant can be less
efficient than from a smaller local power plant. The wear and tear of machinery also
lower fuel efficiency, but this can partly be restored through maintenance and ser-
vicing.
All power plants – except those based on modular generating units in
parallel – show an increase in specific fuel consumption if their output decreases
from maximum output. This is caused by a higher impact of parasitic losses, such
as pump drives, and of system losses such as bearing friction and the blow by of
turbine blades when running at reduced load. By contrast, modular power plants
based on multiple engines can maintain their full load fuel efficiency down to
very low loads by switching off individual generating units. The units that remain
online can always run close to their maximum output. Figure 6.7 gives the typical
fuel efficiencies of different electricity generating techniques versus load. Most
power plants will not run below 50% of nominal output because of technical and
economic restrictions.
Fuel efficiency in combination with fuel price determines the fuel costs expressed
in €cnt/kWh. Since fuel prices largely depend on market forces, it is impossible for
power plant projects to predict the future fuel costs. Figure 6.8 gives examples of
fuel costs for a coal price of 0.24 €cnt/
MJ, a gas price of 0.35 €cnt/MJ, and a Table 6.3.  Some values of power plant effi-
nuclear fuel price of 0.10 €cnt/MJ. It ciencies at nominal output.
should be noted that a large part of the
nuclear fuel price is determined by the Power plant type Fuel efficiency
at nominal output
processing costs and less by the com- Nuclear 33%
modity costs of raw uranium oxide.
Hard coal 40%
The volatility in nuclear fuel costs is,
therefore, relatively small. However, Gas turbine simple cycle 38%

in the case of many new build nuclear Gas turbine combined cycle 55%
power plants, fuel scarcity would drive
Gas engine simple cycle 47%
up the price. Currently, nuclear power
130  Power supply challenges

plants have by far the lowest specific fuel


costs of all fuel-based power plants. But 60
the low gas price of 0.24 €cnt/MJ in the 55
USA in 2013 rendered specific fuel costs GT CC
50

Fuel efficiency (%)


for gas-fuelled plants of close to 1.5 €cnt/ Modular gas engine
45
kWh. For such boundary conditions,
gas-based power plants can even push 40
Coal
nuclear power plants out of the market 35
GT SC
for baseload. 30
Nuclear
A complicating factor in deter- 25
mining fuel costs are the effects on fuel
20
efficiency of rapid load variations and 0 20 40 60 80 100
frequent starts and stops of the gener- Relative power plant output (%)
ating equipment. Steam boilers suffer
from rapid changes in temperature and
Figure 6.7.  Examples of the fuel efficiencies for different
their starting-up and stopping processes
electricity generating techniques as a function of relative
are, therefore, relatively slow. Pre- power output.
heating and a slow ramping up of power
plants consume extra fuel. Combustion
engines and aero-derivative simple-cycle Coal 0.24 €cnt/MJ
Gas 0.35 €cnt/MJ
combustion turbines are not afflicted 4.5 Nuclear fuel 0.10 €cnt/MJ
with this problem. As seen in chapter 3, 4.0
Specific fuel costs (€cnt/kWh)

GT SC
large fractions of wind and solar-based 3.5
generators in a system mean that fast 3.0
Modular gas engine GT CC
ramping up and down and frequent 2.5
starts and stops of fuel-based generators Coal
2.0
become common practice. The intro-
1.5
duction of wind and solar increases, Nuclear
1.0
therefore, the specific fuel costs of the
other power plants. 0.5

0.0
0 20 40 60 80 100
6.3.3. Emission costs
Relative power plant output (%)

In some countries, a fee is charged for


emitting greenhouse gases such as CO2, Figure 6.8.  Examples of the fuel costs for different power-
pollutants such as NOX, SO2, and particu- plant techniques fuel efficiencies as in Figure 6.7, and for given
lates. In other countries, CO2 emitters have boundary conditions of fuel prices.
to purchase emission certificates. There
are methods available that can reduce the emissions of pollutants, but carbon capture
and storage from exhaust gas streams (CCS) is not common practice. The equipment
is expensive, capturing gases consumes additional energy, and there are controversies
about where to store the captured CO2. Proposals for minimising greenhouse gas emis-
sions from fuel use via trade and cap certificates have failed because of an excess of
6.  Costs of producing electricity  131

Table 6.4.  Examples of CO2 emissions depending on power plant type and fuel (nom-
inal conditions).

Power plant type CO2 from fuel Nominal fuel efficiency Specific CO2
emission

g/MJ % g/kWh

Black coal/steam 98 40 882

Lignite/steam 109 36 1090

Gas/simple cycle gas turbine 56 38 530

Gas/comb. cycle gas turbine 56 55 366

Gas/combustion engine 56 47 429

Gas/engine with cogeneration 56 47 + 40 232

certificates, resulting in a very low CO2 price of about 5 €/tonne as of early 2014. Such
a low price is clearly insufficient to finance CCS. A CO2 price of between 60 € and 80 €
per tonne would be needed for making CCS possible.
The CO2 emissions expressed in g/kWh depend on the composition of the fuel
and the fuel efficiency. Table 6.4 gives some examples for steady-state full-load condi-
tions of different power generating techniques.
If the CO2 price would rise to 80 €/tonne, the emission costs for coal-based
power plants would be more than 7 €cnt/kWh. This is due to the high CO2 emis-
sions from coal and the relatively low fuel efficiency of coal power plants. This is
illustrated in Figure 6.9. For a decreasing load, the fuel efficiency decreases and

7
Specific CO2 costs (€cnt/kWh)

Coal
6
CO2
5 80 €/tonne
4
GT SC
Mod gas engine
3 GT CC
2 Coal
GT SC CO2
1 Mod gas engine 20 €/tonne
GT CC
0
0 20 40 60 80 100

Relative power plant output (%)

Figure 6.9.  Costs of CO2 emissions per kWh with two different CO2 price scenarios (fuel
efficiencies as in Figure 6.7).
132  Power supply challenges

therefore the CO2 costs rise. Even for a CO2 price of 20 €/tonne, the CO2 costs
of a coal-fired plant reach a level close to its capital costs as shown in Figure 6.6.
Apparently, the impact of pricing of CO2 emissions is so high that it can easily
increase or decrease the market viability of a generating technique. Gas-based
plants clearly have an advantage because gas releases less CO2 per MJ than coal,
while gas plants also have higher fuel efficiency. Gas turbine combined cycles and
power-plants based on multiple engine-driven generators have the lowest CO2
costs per kWh. Where engines are used in a combined heat and power plant, their
CO2 costs will be halved.

6.3.4. Operation and maintenance costs


Running technical equipment, such as a power plant, causes wear. Turbine blades,
bearings, valves, air filters, lubricating oil and exhaust-gas-cleaning catalysts deterio-
rate during the operation of a power plant. The machinery housing and administra-
tive offices need regular cleaning and maintenance. In addition, power plants need
personnel for operating and supervising the machinery, despite a high degree of auto-
mation in controlling and monitoring the
equipment.
A distinction can be made between 1.4
fixed operation and maintenance
Specific variable O&M costs (€cnt/kWh)

(O&M) costs and variable O&M costs. 1.2


Nuclear
Fixed costs, such as those for personnel
1.0
and insurance, occur independently of
Coal + GTs
the utilisation factor of the generating 0.8
system. By contrast, variable costs are
directly dependent on the utilisation 0.6
factor. Boilers and turbines also experi- Modular gas engine plant
ence additional wear due to stopping, 0.4

starting and rapid load changes. Such


0.2
events are expressed in equivalent run-
ning hours. Reciprocating engines, how- 0,0
ever, can experience frequent starts and 0 20 40 60 80 100
stops without causing additional wear.
Power plant load(%)
Variable maintenance costs are gen-
erally based on running hours without
taking into account the fact that the load Figure 6.10.  An example of variable O&M costs for different
level might affect the wear rate. Variable power plant types depending on the load of the power plant
operational costs are anyhow based only (100% = nominal output).
on the number of running hours, since
running at reduced output does not decrease the required operational activities.
Figure 6.10 is an attempt to relate the variable O&M costs with the load for different
power plant types. The effect of frequent starts and stops on maintenance needs has
6.  Costs of producing electricity  133

not been included in this. Each start and stop of a supercritical boiler or turbine adds
equivalent running hours to a power plant’s maintenance requirement. By contrast,
the maintenance costs per kWh of power plants based on multiple combustion
engine-driven generators do not depend on the load since individual generating
units will be switched off when the load decreases, meaning that the wear factor
consequently ceases. This is another advantage of modular plants in situations where
the need for fuel-based electricity is reduced by a substantial input from renewable
electricity sources.
In real life, the variable O&M costs can differ substantially from those shown
in Figure 6.10. Costs depend on the equipment quality, the fuel quality, as well as
the operating conditions, which can vary from easy to difficult. As an example, poor
quality black coal burns so slowly that boilers suffer from thermal overheating of the
burner grates.
When the utilisation factor of power plants is reduced, either because of poor
market conditions or through switching from full-time production to becoming a
backup plant for renewables, the fixed O&M costs per kWh output increas. Table
6.5 gives an overview of the published average fixed O&M costs. Also, these values
can differ substantially from case to case. A wind turbine with a capacity factor of
20% produces 0.2 ∙ 8760 h = 1752 kWh per year per installed kW. In that case, the
fixed annual O&M costs of 13 €/kW become 1300/1752 = 0.74 €cnt/kWh. A nuclear
power plant with a utilisation factor of 95% has fixed O&M costs of only 0.25 €cnt/
kWh. If that nuclear power plant would run only 4 months of the year at an average
output of 80% during the winter time to cover the lack of electricity production from
solar panels, the fixed O&M costs would be some 12/4 ∙ 100/80 ∙ 0.25 = 0.93 €cnt/
kWh. A modular plant based on gas engines used for peaking and secondary reserves

Table 6.5.  Examples of fixed O&M costs for different power-plant techniques.

Power plant type Fixed O&M costs per year

€ per kW installed capacity

Hard-coal/steam 16

Lignite/steam 18

Nuclear/steam 21

Gas GT comb. cycle 10

Gas engine modular 15

Wind turbine 13

Solar PV 8

Hydro 8
134  Power supply challenges

only might have a utilisation factor of 30%, resulting in fixed O&M costs of 15 ∙ 100/
(0.3 ∙ 8760) = 0.57 €cnt/kWh.

6.4. The total costs of producing electricity


It would be great if the previous sections of this chapter would have provided a simple
table revealing the costs per kWh for producing electricity with the different techniques.
That appears to be impossible since the boundary conditions for each technique differ,
depending on the costs of financing, fuel, construction, emissions, plus the utilisation
and load factors. A fully depreciated nuclear power plant can produce electricity at very
low marginal costs, but as soon as the capital costs are taken into account and the utilisa-
tion factor is low, the kWh costs can become very high. A coal-fired power plant might
beat a gas-fuelled power plant in kWh costs, but if the price of CO2 emissions rises
substantially, coal-based plants cannot compete anymore.
Figure 6.11 gives the total kWh cost for a number of generating methods for the
boundary conditions given in figures 6.12, 6.13 and tables 6.4 and 6.5. It is important
to know that any additional costs, such as the need for reserve power and frequency

20
18
Total kWh costs (€cts/kWh)

16
14
12
10
8
6
4
2
0
Hard Nuclear GTCC Mod. gas Onshore Offshore Solar Hydro
coal engine wind wind
Costs:
Total 6.06 5.34 4.68 4.72 9.34 12.17 19.61 1.4
CO2 1.6 0 0.73 0.85 0 0 0 0
Fixed O&M 0.2 0.25 0.13 0.19 0.74 0.57 0.61 0.1
Var. O&M 0.5 0.6 0.76 0.5 0.4 0.5 0.6 0.4
Fuel 2.16 1.09 2.3 2.68 0 0 0 0
Capital 1.6 3.4 0.76 0.5 8.2 11.1 18.4 0.9

Figure 6.11.  Total production costs of electricity where the CO2 price is 20 €/tonne, the
utilisation factor is 90% (except wind and solar) and the plant is running at full output
[coal price 0.24 €cnt/MJ, gas price 0.35 €cnt/MJ, nuclear fuel 0.10 €cnt/MJ, other condi-
tions from table 6.4 and 6.5 and Figure 6.10].
6.  Costs of producing electricity  135

20
Total kWh costs (€cts/kWh) 18

16

14

12

10

0
Hard Nuclear GTCC Gas Onshore Offshore Solar Hydro
Costs: coal engines wind wind
Total 10.86 5.34 6.85 7.27 9.34 12.17 19.61 1.4
CO2 6.4 0 2.9 3.4 0 0 0 0
Fixed O&M 0.2 0.25 0.13 0.19 0.74 0.57 0.61 0.1
Var. O&M 0.5 0.6 0.76 0.5 0.4 0.5 0.6 0.4
Fuel 2.16 1.09 2.3 2.68 0 0 0 0
Capital 1.6 3.4 0.76 0.5 8.2 11.1 18.4 0.9

Figure 6.12.  Production costs per kWh with a CO2 price of 80 €/tonne [other boundary
conditions as in Figure 6.11.

control, are not included in the production costs. The costs given are purely for the
electric energy. For the conditions applying in Figure 6.11, the kWh costs of the fuel-
based power plants are around 5 €cnt/kWh and mutual competition is a good pos-
sibility. Hydropower beats all other sources in costs and that is why countries blessed
with such resources are very fortunate. Renewable energy based on wind and solar
radiation is costly because of low capacity factor. Again, the costs for backup capacity
for wind and solar-based electricity sources have not been taken into account.
Fuel efficiencies, O&M costs, and the price of equipment for fuel-based gen-
erators and hydro-electric plants are not expected to change drastically in the near
future. However, major uncertainties exist for the costs of CO2 emissions and fuel.
Should the price of coal be 0.15 €cnt/MJ instead of 0.24 €cnt/MJ, the costs per kWh
of the coal-based plant would drop from 6.06 €cnt/kWh to 5.25 €cnt/kWh. If the
price of CO2 is 5 €/tonne instead of 20 €/tonne, the cost of coal-based electricity is
only 4 €cnt per kWh instead of 6.06 €cnt/kWh. If the gas price would be 0.7 €cnt/
MJ instead of 0.35€cnt/MJ, which is currently the case in Europe, combined cycle
and modular gas engine-based plants would have electricity costs of about 7 €cnt/
kWh instead of around 4.7 €cnt/kWh.
136  Power supply challenges

Marginal kWh costs 5

4
(€cts/kWh)

0
Hard Nuclear GTCC Gas Onshore Offshore Solar Hydro
Costs: coal engines wind wind
Marginal 4.16 1.69 3.53 4.05 0.4 0.5 0.6 0.4
CO2 1.6 0 0.73 0.85 0 0 0 0
Var. O&M 0.5 0.6 0.5 0.5 0.4 0.5 0.6 0.4
Fuel 2.16 1.09 2.3 2.68 0 0 0 0

Figure 6.13.  Marginal kWh costs for a utilisation factor of 90% and a nominal load for
the boundary conditions of Figure 6.11.

These changes in boundary conditions reflect the situation in Western Europe in


2013 and explain why coal-fired power plants had the largest market share in that
year. In North America, the low gas price of close to 0.24 €cnt/MJ pushed coal-
fired power plants out of the market. Therefore, although the determination of the
competitiveness of power plants is difficult, modifying Figure 6.11 to project-specific
boundary conditions can be very helpful when estimating the costs of producing
electricity.
If CO2 costs would rise to 80 €/tonne, coal-fired power plants would be com-
pletely pushed out of the market, even with favourable conditions, i.e. a 90% utilisa-
tion factor and running at full output. This is illustrated by Figure 6.12. Therefore,
the consequences of high CO2 emission charges for countries depending on coal-
based power plants will be severe. One should not make the mistake of believing
that, based on Figure 6.12, on-shore wind turbines are becoming competitive with
coal-fired power plants. Wind turbines always need backup capacity. They can only
push other generators from the market in times of sufficient wind.
In open electricity production markets where power plants have to compete by
offering electrical energy in short-time intervals, marginal production costs are often
used. The competing power plants are present anyhow, and if they can sell electricity
at a price higher than their marginal costs they receive at least an income stream.
The difference between production costs and market price might not be sufficient
to cover the fixed costs, but getting some income is always better than no income
at all. Ultimately however, a power plant should recover the total costs of producing
electricity.
6.  Costs of producing electricity  137

Figure 6.13 gives only the marginal costs of producing electricity for the
boundary conditions of Figure 6.11. In this case, the renewable electricity sources
beat all other power generators. Fully depreciated power plants, for example many
old coal plants, have a substantial advantage over new gas-based power plants.
And again, if the coal price would be just 0.15 €cnt/MJ instead of 0.24 €cnt/MJ,
the CO2 price 5 €cnt/tonne instead of 20 €/tonne, and the gas price 0.7 €cnt/MJ
instead of 0.35 €cnt, the gas-based plants would be fully pushed from the market.
The coal-based marginal electrical energy costs would be only 2.4 €cnt/kWh,
while the gas-based marginal costs would be about 5.7 €cnt/kWh. Some owners
of power plants even leave the maintenance costs out when offering electricity to
the market. They will run their power plant until the moment a major overhaul
becomes necessary and then decide if it is worthwhile to carry out the necessary
maintenance actions or to close down the power plant. In free electricity markets,
everything is possible.
The story this far illustrates the difficulty power plant investors face in making
the proper choice for a power plant. The situation gets even more complicated when
many renewable electricity sources are taken into the system. A utilisation factor

16
Total kWh price (€cts/kWh)

14

12

10

0
Hard Nuclear GTCC Mod gas Large scale
Costs: coal engine hydro
Total 11.73 15.95 7.54 6.64 4.31
CO2 1.76 0 0.83 0.86 0
Fixed O&M 0.76 1 0.48 0.71 0.34
Var. O&M 0.83 1 0.83 0.52 0.6
Fuel 2.38 1.2 2.55 2.68 0
Capital 6 12.75 2.85 1.87 3.37

Figure 6.14.  kWh costs where the utilisation factor is 24% and the plant is running at a
generator output of 60% [other boundary conditions as in Figure 6.11.
138  Power supply challenges

of 90% and running only at 100% load will seldom be possible anymore for power
plants.
Chapter 3 shows that even without renewable electricity sources, the average
capacity factor of a real power-plant portfolio is hardly higher than 50%. The Irish
example shows that when wind energy provided 16% of the electrical energy, the
utilisation factor of the other power plants was only 39%. Without wind power, the
utilisation factor would have been about 47%. If the Irish wind-based capacity were
to double, the utilisation factor of the power plants drops to 32%. Yet, the power
plants need to be able to supply peak demand as long as no energy storage system
is available for covering a lack of wind output. The German case shows that solar
panels there have a lower capacity factor than wind turbines. Therefore, the backup
capacity for solar panels generally has a higher utilisation factor than the backup
capacity for wind. However, the daily variations are larger for solar than for wind
energy. This makes investing in power plants even more complicated.
In the hypothetical case that a power plant will only run at a fixed load of 60%
and be online for only 40% of the time, its utilisation factor is only 0.6 ∙ 40 = 24%.
The electricity production costs will then be much higher than when running 90%
of the time at 100% load. The production costs per kWh for the 60% load case are
given in Figure 6.14. The low utilisation factor means that those power plants that
require high investments per kW cannot compete anymore because of the burden

End user electricity price (€cnt/kWh), year 2011

30
Households
Industry
25

20

15

10

0
Belgium

Finland

France

Germany

Italy

Netherlands

Norway

Poland

Spain

Sweden

Turkey

United Kingdom

USA

China
Denmark

Figure 6.15.  End user electricity prices in different countries (sources: EIA, Eurelectric,
Eurostat, www.europe.eu).
6.  Costs of producing electricity  139

of the capital costs. This reveals that new nuclear power


and coal plants have no future in countries such as Ire-
land and Germany because they have a lot of renewable Additional
charges 0.6
VAT
15.9
energy sources. If the price of CO2 were to quadruple to Measurement Energy
levy 2.5 24.1
80 €/tonne compared with the 20€/tonne in Figure 6.14, Concession
backup electricity from coal would cost 17 €cnt/kWh. levy 6.4
Electricity use
Lower coal prices will not really help in this case. levy 7.9 Grid charge
20.6
In reality, power plants with a utilisation factor of only Retail
8.2
24% will never run constantly at 60% load. The increasing Renewables
levy 13.8
need to fulfil the tasks of frequency control, load fol-
lowing, reserve power, and forecast-error compensation
means that their output will change all the time. Nuclear Figure 6.16.  Percentages of the different
and coal-based power plants are less suited to changing items making up the 26.1 €cnt/kWh price of
output because they suffer if steam conditions vary. That electricity for households in Germany (Data
from year 2012).
also applies for the steam part of gas turbine combined
cycles. Therefore, their variable maintenance costs will be
even higher than indicated in Figure 6.14 while their fuel efficiency will be lower.
Currently, some owners of combined cycle power plants disconnect the steam
cycle from the gas turbine. This is to avoid wearing of the heat recovery steam gen-
erator and steam turbine, and to create more flexibility. The steam part of the cycle
responds relatively slowly to changes in output demand.

6.5. The electricity price for consumers


As seen in chapter 1, the price of a product normally differs from the costs to produce
it. Generally, the overall production costs of electricity vary between 2 €cnt/kWh for
existing hydro-electric power plants as a minimum, and 20 €cnt/kWh from solar panels
as a maximum. Occasionally however, even higher costs might occur in rural locations
where small petrol-fuelled generators are the only source of electricity.
Prices for electricity differ considerably between countries, even in an area
such as the European Union. Figure 6.15 gives the approximate electricity prices
in a number of countries. A Danish household pays twice as much per kWh than
a domestic customer in France or Finland, for example. Germany ranks second in
kWh price for domestic users. Denmark and Germany are the EU countries with the
highest share of renewable electricity sources. Industry in the USA pays the lowest
kWh rates. This certainly enhances the competitiveness of energy-intensive indus-
tries.
Domestic consumers generally pay considerably more per kWh than industrial
users. This is partly due to higher distribution costs for homes than for large con-
sumers. However, levies, taxes and distribution charges play a significant role here.
They can result in domestic customers paying four times as much as industry.
Figure 6.16 shows the composition of the price that households paid in Germany
in 2012. The domestic retail price is 26.1 €cnt/ kWh, so it is evident that solar
140  Power supply challenges

panels producing electricity for 20 €cnt/kWh are competitive. However, it appears


that only 6.3 €cnt/kWh of the retail price of electricity is for the energy. Competi-
tion between electricity producers has, therefore, a close to negligible effect on the
price that a domestic consumer pays in countries such as Germany, Denmark, The
Netherlands, Italy, Spain and Sweden. In these countries, domestic electricity use is
heavily charged with levies and taxes. High electricity prices for industry can result
in energy-intensive companies leaving for cheaper regions.

6.6. Discussion regarding electricity production costs


At first sight, building up an electricity generation portfolio based on minimum costs
seems simple. Techniques with low fuel prices combined with relatively high capital
costs are the most suitable for baseload. Techniques with low capital investments but
higher fuel costs are the best option for intermediate and peak load. Hydropower based
on a sufficient water supply in reservoirs is the cheapest option for all applications but
is not available in most countries.
However, the basic production costs of electricity are not the only issue here.
Electricity supply systems also need contingency reserves, frequency-control
capacity, load-following capacity, reactive power supply, and forecasting-error
reserves. These ancillary services cost money. In traditional vertically integrated
power supply systems, they were an integral part of the tasks of a power plant.
Running a certain fraction of the online power plants below their nominal output
could generally provide these services. Even in competitive power markets, system
operators may charge electricity producers with ancillary services as part of the
deal to supply energy.
In some markets, power suppliers can receive compensation for offering ancillary
services. But not all generating techniques have similar capabilities for providing
these services. With electricity sources of intermittent output, the need for ancillary
services, including fast ramping up and down and even energy storage, drastically
increases. In open electricity markets, such services should certainly be rewarded
financially.
It is quite a challenge to find the economically and technically optimum for
building up a generating portfolio as conditions can change in an unpredictable way.
However, this chapter has shown that hydropower and gas-based power tend to offer
the cheapest solutions when significant levels of flexibility are required. Power plants
based on multiple units in parallel have the additional advantage of low maintenance
costs and high fuel efficiency, even at low loads.

6.7. Conclusions
This chapter has shown that the costs of producing electricity with generators oper-
ating in competitive markets depend heavily on unpredictable boundary conditions.
6.  Costs of producing electricity  141

This makes it extremely difficult to predict the profitability of existing power plants,
and especially that of future power plants. Political interference, fluctuating fuel prices,
and the increasing amount of subsidised renewable electricity sources heavily affect the
ultimate cost of electricity.
Despite the many uncertainties, new power plants are definitively needed over
the coming decades because of the crucial role of electricity in the economy. Renew-
ables alone cannot effectively cover the total electricity demand in the foreseeable
future. Flexibility is the key to coping with the uncertainties in power generation
systems. Where hydropower is not available, the best means for achieving flexibility
appears to be modular gas power plants based on combustion engine technology.
7 Future power
supply systems
Future power supply systems require first and foremost flexible power generation
to ensure system reliability and optimisation at the lowest possible cost. Flexible
generation enables sustainability by allowing large amounts of renewable energy –
while maintaining grid stability. Agile, fast-reacting generating units offer optimum
solutions for the many challenges that power supply systems face. This chapter
shows how power plant portfolios can be optimized with flexible generating units
in different supply systems.
144  Power supply challenges

7.1. The road towards an optimum power supply system


Traditionally, power plants were built to last for at least forty years. However, radical
changes caused by the introduction of subsidised renewable electricity sources and by
open competition in energy markets can make even brand new power plants obsolete.
In 2013, a one-year-old combined-cycle plant near Rotterdam was dismantled and sold
in parts to regions with better boundary conditions. Major power suppliers have been
confronted with fully unexpected circumstances.
The previous chapters have shown that maintaining stability in electricity supply
systems is becoming increasingly complicated. Planning and designing a generation
and transmission portfolio has never before been so difficult. The crucial role of elec-
tricity in the economy requires a highly reliable supply of quality electricity with, at
the same time, minimum costs. The large-scale introduction of renewable electricity
sources with variable and uncertain output requires completely new approaches.
Flexibility in power-plant output, as well as fast ramping up and down, is much more
needed now than in the past. Stakeholders also talk about the need for expanding
grids, demand-side response and energy storage. Long-term profitability is needed
but is very hard to predict because of continuously changing support schemes for
renewables, emission reduction techniques and the charges for emission rights.
This book cannot give a single solution for achieving the optimum power supply
system, since local boundary conditions determine what the best option is. However,
this chapter will show that flexible and agile generators substantially help to increase
reliability and stability in very different power systems. Furthermore, they appear to
do this with good fuel efficiency and minimum investment costs.

30 minute interval demand year 2012,


50 Hertz transmission
15 000
12 500
10 000
Power (MW)

7 500
5 000
2 500
0
0 10 20 30 40 50
Weeks of the year 2012

Figure 7.1.  Power demand pattern in the German 50 Hertz Transmission area during
the year 2012.
7.  Future power supply systems  145

7.2. An optimised generating Demand distribution year 2012,


portfolio without renewables 50 Hertz transmission
15 000
12 500
Traditionally, in times when the flow of

Power (MW)
10 000
electrical energy was almost entirely one-
7 500
directional, from the power plant to the 5 000
consumers, electricity demand patterns, 2 500
and consequently electricity production 0
patterns, were quite predictable. The effect 0 10 20 30 40 50 60 70 80 90 100

of daily activity patterns, weekends, public % of the time

holidays and the weather on electricity


consumption is always clearly distinguish- Figure 7.2.  Power demand distribution in the German 50
able in power demand patterns and is, Hertz Transmission area during the year 2012.
therefore, easy to forecast. Figure 7.1 gives
a typical example of a power demand pattern. It is based on data at 30 minute inter-
vals for the 50Hertz Transmission System Operator area in Germany for the year 2012.
Apparently, demand there was never higher than 14 GW and never lower than 5 GW.
The data points of figure 7.1 have also been transferred into the power demand distribu-
tion curve shown in figure 7.2.
We will now make a hypothetical case of creating a power-plant portfolio that
can fulfil the power demand as given in figure 7.1 in an optimised way. Chapter
3 shows that in reality, extensive wind and solar based capacity is present in the
50Hertz TSO area, but that will be presumed not to exist in this example. In addi-
tion, transmission and distribution losses will be neglected, since those losses only
act as extra load. Power plants will be needed to cover the baseload, intermediate
load, and the peak load. In addition, generating capacity is needed for primary,
secondary and tertiary reserves, as well as for plants that
are undergoing maintenance. Figure 7.3 summarises the
Replacement for tertiary reserves
different modes of generating capacity that have to be
Tertiary reserves
covered.
Secondary reserves
Power demand in the 50Hertz TSO area is always
Primary reserves
higher than 7 GW, apart from very few downward excur-
sions. In an idealised situation, ten power plants each of 700 Maintenance reserves

MW can carry this baseload. However, power plants require Peak load
regular maintenance so that an additional reserve plant of
700 MW is needed as backup to guarantee sufficient avail- Intermediate load
ability. In addition, sufficient primary reserves have to be
available in case one or more power plants are down. One
option for this is to run 11 power plants, each of a 700 MW Base load
nominal capacity at a load of 636 MW, to supply the 7 GW.
Running at 91% load has no severe negative consequences
for fuel efficiency and increases the capital costs per kWh Figure 7.3.  The different modes of gener-
by only 10%. When one of the baseload plants fails, the ating capacity in which has to be invested
146  Power supply challenges

remaining ones have to ramp up to 700 MW within 30 seconds when following the
typical Continental European rule for primary reserves.
Secondary reserves are needed to add extra power to the system so as to
restore the frequency to 50 Hz so that the primary reserves can return to the pre-
contingency output. Therefore, 636 MW should be allocated for secondary reserves.
An alternative is to add another 700 MW power plant to the baseload portfolio and
run each of the twelve power plants in parallel at 583 MW, equalling 83% load. A
much better solution, however, is to use a quick-start non-spinning power plant of
636 MW. Engine-based generating units can deliver full output within some five
minutes after the start command, as has been shown in chapter 3. The benefit of this
is that the eleven online baseload plants that satisfy the 7 GW demand can remain
running at 91% load with, therefore, better fuel efficiency than when running at 83%
load. Moreover, a non-spinning power plant does not accumulate wear while waiting
to perform its task as a secondary reserves provider. In addition, such a power plant
requires a considerably lower investment than a dedicated baseload power plant, (see
figure 6.6). The ultimate output reliability of a non-spinning reserve power plant
based on, for instance, 25 identical units in parallel, is much higher than that of a
single unit.
In addition to the secondary reserves, another 636 MW of non-spinning tertiary
reserves has to be present to release the secondary reserves following a contingency.
Typical steam-based power plants can never provide full output from standstill

Figure 7.4.  Photograph of an agile engine-based multiple-unit power plant


7.  Future power supply systems  147

700 MW 10.36 GW

700 MW 61.5 TWh 400 MW 4.4 GW 200 MW 2.2 GW

700 MW 700 MW 400 MW 18 TWh 200 MW 6 TWh


replace
reserve
700 MW 400 MW 200 MW

700 MW 700 MW 400 MW 200 MW


mainten.
reserve
700 MW 400 MW 200 MW

700 MW 630 MW 400 MW 200 MW


agile
tertiary.
700 MW reserve 400 MW 200 MW

700 MW 400 MW 200 MW


700 MW
700 MW agile 400 MW 400 MW 200 MW 200 MW
second. mainten. mainten.
reserve reserve reserve
700 MW 400 MW 200 MW

Base load generators Intermediate load generators Peak load generators

Figure 7.5.  An optimum power plant portfolio to satisfy the demand pattern of figure 7.1.

within the required time span of one hour, as required for tertiary reserves. More
agile power plants offer a solution in this case.
To fulfil the power demand as shown in figure 7.1, intermediate generation is
also needed in the demand range between 7 GW and 11 GW. This requires another
4 GW of capacity to be regularly is online. Intermediate demand is characterised
by power ramping up in the morning and down in the late evening, especially on
weekdays. The 4 GW can be supplied by 10 units of 400 MW each. The reserves for
intermediate power generation can be warranted by the reserves allocated for the
baseload plants. Yet, a reserve plant for intermediate power is needed because the
other generators need regular maintenance.
Peak demand, occurring less than 30% of the time, requires 2 GW of agile power
plants plus a reserve unit to cover maintenance. Again, in a simple approach, 11
plants each of 200 MW are required. One of these power plants can also be used for
compensating for errors in forecasting demand. Figure 7.5 gives the resulting power
plant portfolio to cover the demand shown in figure 7.1.
In 2012, the total electric energy demand was 85.5 TWh in the 50Hertz TSO
region. With the power plant portfolio proposed in the previous paragraph, the fif-
teen plants dedicated to baseload including the reserves would produce 7 GW · 7884
h = 61.5 TWh from an installed capacity of 13 ∙ 700 MW + 2 ∙ 636= 10.37 GW. This
yields a utilisation factor of 61.5 TWh/(10.37 GW· 8784 h) · 100% = 67.5% for the
baseload plants.
148  Power supply challenges

Figure 7.2 shows that the intermediate load, between 7 GW and 11 GW, will be
needed about 51 % of the time. That can be converted into 4 GW · 8784/2 h = 18
TWh. Therefore, the 4.4 GW of the eleven intermediate load plants would have a
utilisation factor of 46.6%. The 2.2 GW of the eleven peaking units would produce
the remaining 85.5 – (61.5 +18) = 6 TWh with a utilisation factor of 31.0%. The
combined portfolio has a utilisation
factor of 57.4%. Optimised generation portfolio
In real life, the utilisation factor of a 100
power plant portfolio is generally lower 100

than the 57.4% of this example. The main 80


61
67.8
57.4
reason is that many countries have an 60 46.6
annual demand pattern with a relatively 40
26
31
lower baseload than in figures 7.1 and 20 13
7.2. In countries with hot climates, the 0
baseload in the mild seasons is lower Total Base load Interm. load Peak load
than in hot seasons. In cold countries, Installed capacity (%) Utilisation factor (%)
baseload is highest in winter.
Recently, in countries with significant Figure 7.6.  Subdivision of installed generating capacity with
levels of wind and solar-based generation, the connected utilisation factor
such as Denmark (DK), Germany (DE),
Italy (IT) and Spain (ES), the utilisation factor of the fuel-based power plants has
decreased drastically. In addition, new-built power plants with high fuel efficiency
often take on the role of the old power plants, while these old plants remain and act
to provide reserve capacity.
Figure 7.7 gives the utilisation factor of the power plants of 22 EU countries in
the year 2012. The highest value is almost 50% and the lowest is 13.7%. The average
utilisation factor of the power plants for the 22 countries is only 37.3%, as shown by
the red line. This shows that the example in figure 7.5 is clearly an idealised case.

Power plant utilisation factor 22 EU countries, year 2012


50
Utilisation factor (%)

40
30
20
10
0
AT BE CZ DE DK ES FI FR GB GR HU IE IT LT LU LV MT NL PL PT RO SE

Figure 7.7.  The utilisation factor of the power plant portfolio of 22 EU countries in 2012
(data from Eurelectric).
7.  Future power supply systems  149

For selecting the best generating techniques for the power plant portfolio shown
in figure 7.5, chapter 6 of this book clearly shows the best options from an economic
point of view. The thirteen 700 MW baseload power plants in our example could
be nuclear or coal-fired plants. The remaining two 636 MW plants in the baseload
portfolio have to be more agile and flexible, since they have to provide the secondary
and tertiary contingency reserves. The intermediate plants could be gas turbine
combined cycles and the peaking units could be simple-cycle gas engines and gas
turbines. Hydropower, if abundantly available, would be the first choice for all; base-
load, intermediate load, as well as peak load.
The total costs of producing electricity with the power plant portfolio proposed
in figure 7.5 can be determined with the methods shown in chapter 6. Planners and
economic analysts for power plants have traditionally followed the methodology
described here and used investment costs and variable boundary conditions for such
items as fuel prices to determine the competitiveness of a typical power plant type.
In the old days, the only option for secondary reserves was spinning power plants.
The current availability of agile power plants that offer non-spinning reserves means
that fuel consumption, maintenance costs and investment costs can be decreased.

7.3. An optimised power plant portfolio design with renewable energy


sources
When renewable electricity sources are added to the system, power supply system plan-
ners cannot simply use a power demand distribution curve such as that in figure 7.2
anymore. Such an approach was only possible for a highly predictable load pattern for
the power plants. This will be shown by using again data from the 50 Hertz TSO area
in Germany.
Germany is pioneering the large-scale introduction of renewable electricity
sources. In 2012, the 50Hertz TSO area had on average 12 GW of wind power
installed with a capacity factor of 17.6 %,
and an average 8 GW of solar PV with a
Average values, 50 Hertz TSO, year 2012
capacity factor of 9.5 %. These two renew-
60
able sources produced 18.5 TWh and 5.1
50
TWh of electricity respectively. The total
40
electric energy demand for the area was
30
85.5 TWh in 2012. The wind turbines
20
and solar panels produced, therefore,
10
27.6% of the electric energy demand in
0
the 50Hertz TSO region. Wind Solar Others
Sometimes high winds occur on
sunny days. Then the output from wind- Capacity (GW) Production (TWh)
turbines can be as high as 90% of the
installed wind-based capacity, while solar- Figure 7.8.  Installed generation capacities and their output,
panel output can be 80% of the installed 50 Hertz TSO.
150  Power supply challenges

solar-based capacity. The total output


from the renewable sources is thus 0.9 ∙ Demand distribution year 2012, 50 Hertz transmission
12 + 0.8 ∙ 8 = 17.2 GW. This output even 15 000

exceeds the 14 GW maximum demand 12 500


Total demand
of the 50Hertz TSO area. Chapter 3 10 000

Power (MW)
shows that such a situation occurred 7 500
only a few times in 2012. However, the Demand minus
5 000
expansion of renewables capacity has renewable contribution
2 500
not ended. Therefore, high outputs from
renewable electricity sources can easily 0
push all other generators from the grid if –2 500
their output is not curtailed, exported or 0 10 20 30 40 50 60 70 80 90 100

stored. % of the time

The real-life distribution curve of that


part of the power demand that had to be Figure 7.9.  Distribution curves of total power demand and
delivered by electricity sources other than demand minus renewable contributions in the German 50Hertz
the renewables in the German 50Hertz TSO region during the year 2012 . In the case of no imports/
exports, power plants have to supply the power requirement
TSO region in 2012 is given by the according to the thick red curve.
thick red line in figure 7.9. These other
electricity sources can be power plants,
imports or energy storage. The thick red line has been determined by subtracting the
half-hourly output from wind and solar from the half-hourly power demand curve.
The thick red curve indicates that a substantial load of some 5 GW remained for
power plants during 80 % of the time. At first sight this might offer opportunities for
typical baseload power plants. However, the wind turbines and solar panels create
much volatility in the system. Figure 7.9 illustrates that for some 20% of the time, situ-
ations occurred that required power plants to produce less than 5 GW. A real steady
baseload was no longer present. Relying solely on a power demand distribution curve
is, therefore, inadequate for determining the required power plant portfolio where
there is much wind and solar-based power in the system.
Nevertheless, interesting conclusions can be drawn from the thick red line in
figure 7.9. For almost 2% of the time, the power demand from power plants was less
than 1 GW, while for almost 1% of the time the output from renewables exceeded
even the total electricity demand. Yet, there were times that solar and wind genera-
tion had close to zero output so that the power plants combined with imports still
had to be able to cover peak demand. This can be seen from the overlap of the
dashed power demand curve and the thick red line in the region of peak demand.
Figure 7.9 further shows that power plants plus imports had to cover electricity
demand higher than 10 GW only during 10% of the time. If the renewable genera-
tors had not been present, this would have happened during some 40% of the time.
Because of the very short time during which power plants have to deliver peak
demand of between 12 and 14 GW, peak-demand reduction via incentives for cus-
tomers (demand side response) might be a good option here.
7.  Future power supply systems  151

Demand minus renewables output (30 minute interval data)

15 000

12 500

10 000
Power (MW)

7 500

5 000

2 500

–2 500
0 10 20 30 40 50
Weeks of the year 2012

Figure 7.10.  Power demand minus the contribution from renewable energy sources
contribution in the 50Hertz TSO region during the year 2012

The consequence of the wind turbines and solar panels in the 50Hertz TSO area
in Germany is a substantial variability in the difference between total power demand
and the contribution from the renewable sources. This difference, i.e. the amount of
electricity demand to be covered by power plants and possibly imports, is shown in
figure 7.10. Comparing figure 7.10 with figure 7.1 reveals that designing a power plant
portfolio is now much more complicated than before the introduction of renewables.
From weeks 4 to 6, renewables hardly contributed towards the satisfying of demand.
For example, in the beginning and the end of the year, as well as during week 13,
their output sometimes exceeded power demand. Baseload power plants could still be
applied occasionally up to a capacity of 5 GW, but they had to be frequently switched
off if all the renewable electricity was to be accommodated in the area.
A major difference between the patterns in power demand from power plants
in figures 7.1 and 7.10 is the much higher variability in figure 7.10. In week 3, 2012,
an example of a rapid increase in the required power supply from power plants can
be seen.
Figure 7.11 is a sample of the relevant data in a narrow time span. Wind turbine
output decreased rapidly from 5 GW to 2 GW in the morning, while power demand
increased substantially. Power plants had to add some 6 GW in output within a time
frame of 5 hours from a starting point of 4.5 GW. During a period of two hours,
the required increase was even 1.5 GW per hour. For nuclear and coal-based power
plants, such increases in output after a length of time at standstill are close to impos-
sible. The issue is also that for this occurrence, the initial demand for power from the
power plants was quite low. If steam-based power plants had to cover the balance, it
would only have been possible if they were running at a drastically reduced output
152  Power supply challenges

with much steam throttling at the start


Sample of the effect of drastic renewable output reduction
of the occurrence. on required power plant output in week 3, 2012
It should also be mentioned that the
10

Power supply from non renewables


precise forecasting of substantial changes
in wind-related power is not easy. A shift 9
of half an hour between a predicted 8
change in wind strength and the reality
7
can easily occur. The dashed curves in

(GW)
figure 7.11 show that a 30 minutes earlier 6
or later than predicted change in wind
5
power can make more than 1 GW differ- Demand minus renewable output
ence in the required power plant output. 4 Renewable decrease 30 minutes earlier
Renewable decrease 30 minutes later
The increased variability and higher fore- 3
casting errors mean that power plants 0 50 100 150 200 250 300 350
have to be able to ramp up much faster Time (minutes)
than forecasted or delay their increase in
power output. This is not easy for typical Figure 7.11.  A sample of an occurrence event in the 50 Hertz
baseload power plants. TSO area in Germany, when the wind turbine output dropped
Ramping up and down the output drastically while power demand increased
from power plants is becoming a serious
issue where there are many wind turbines and solar panels in the system. In partic-
ular, it is the relative change in output, i.e. the required change in output divided by
the actual output, that causes difficulties. As an example, ramping up within half an
hour from an initial total combined power plant output of 3 GW to 6 GW is much
more difficult for a portfolio of steam-based plants than ramping up from 9 GW to
12 GW.
Figure 7.12 shows that where there is no wind and solar based power supply in
the system, the relative change in demand from power plants per 30 minutes is gen-
erally lower than + or –15% of their output. With renewable sources in the system,
the change lies most of the time within + or –50% per 30 minutes, a factor three
higher than in the case of no renewables. As a consequence, power plants have to be
much more flexible and agile than in the past.
It should be mentioned here that the relative changes in demand minus renew-
ables shown in figure 7.12 have been calculated based on the presumption that power
plants always provided at least 1 GW of the power demand, even though the renew-
ables sometimes produced more than the total demand. If this presumed minimum
limit of 1 GW would not be applied for the diagram, there would appear occasional
excursions to very high percentages in the relative change in power demand minus
that of renewables. As a result, the figure would become unreadable. However, in
reality the relative changes in load for the power plants will occasionally be even
much higher than those indicated in figure 7.12.
It is clear that a power plant portfolio having significant baseload capacity, as
shown in figure 7.5, is not the proper option where there is a substantial amount of
7.  Future power supply systems  153

100
Relative changes in demand minus

75
renewables per 30 min (%)

50

25

–25

–50

–75

–100
–20 –15 –10 –5 0 5 10 15 20
Relative changes in power demand per 30 min (%)

Figure 7.12.  Comparison of ramping-up and ramping-down rates in power output from
non- renewables with and without renewable sources in the system (based on data from
50 Hertz TSO, year 2012)

power based on wind and solar radiation in the system. In this example from the 50
Hertz TSO area in Germany, only a moderate 27.6% of the required electric energy
came from wind and solar. The intention in Germany and in many other countries
too, is to opt for much higher percentages of renewables, meaning that the typical
baseload for power plants will no longer exist. Consequently, conventional nuclear
and coal-based power plants are ultimately becoming obsolete.

7.3.1. Curtailment or storage of peaks in renewable output


Another point is what to do when the power output from renewables exceeds the power
demand. Exporting temporary excess electrical energy to neighbouring areas is one pos-
sibility. The marginal production costs of electricity from wind turbines and solar panels
is very low, less than 0.5 €cnt/kWh, as has been shown in chapter 6. When wind energy
is exported for such a low price, the power plants in the importing areas cannot compete
with it. As a result, they also will experience higher fluctuations in output and end up
with a lower utilisation factor.
The storage of excessive renewable output is another possibility. A close inspec-
tion of figure 7.9 reveals, however, that the actual energy content of the occasional
excessive output from renewables is not yet very high. If the output of the renew-
able sources were to be curtailed so that at least 1 GW of dispatchable power plants
remained in the system, only 139 GWh of renewable electricity would be rejected in
an entire year. If a minimum dispatchable power of 2.5 GW was chosen, 545 GWh
154  Power supply challenges

Table 7.1.  The effect of curtailing renewable power output on rejected renewable output
(year 2012 data).

Fraction of renewable output


Minimum dispatchable power
Renewable energy rejected rejected

2.5 GW 545 GWh 2.3%

1 GW 139 GWh 0.6%

from the renewables would have been rejected. 139 GWh is only 0.6% of the 23.6
TWh of electric energy produced by wind turbines and solar panels in the 50Hertz
TSO area in 2012. The 545 GWh is still only 2.3%. That’s why policy makers and
decision makers really have to consider if it is worthwhile capturing the occasional
high outputs from renewable sources.
Storage facilities for capturing the excess renewable energy in the case of the 2.5
GW minimum dispatchable power would need an input power capacity of 2.5 + 2
= 4.5 GW, since the electricity production from renewables occasionally exceeded
demand by 2 GW. Since the excess energy to be stored is 521 GWh, a utilisation
factor of 521 GWh/(4.5 GW ∙ 8784 h) ∙ 100% = 1.3% would result for the storage
system. An investment in batteries, pumped hydro, compressed air, or hydrogen
for such a low utilisation factor will never be economic. Even an investment in new
transmission lines to export the excess electric energy would not be economic.
Exporting in combination with storage of the excess energy might be an option if
sufficient existing transmission capacity is available, and if storage capacity could be
offered by already existing hydro capacity from neighbouring Norway, for example.

7.3.2. Converting excess electric energy into heat is a good solution


In the future, much more solar and wind based generating capacity will be installed, so
electricity production from renewables will often substantially exceed demand. Mea-
sures have to be taken to deal with the bulk of the excess energy, although it is better
again to curtail the high peaks that occur only sporadically.
The cheapest option for accommodating and utilising excess electricity appears
to be the installation of electric heating coils in gas-fired hot water boilers, and in the

Table 7.2.  Heat and electricity demand in Germany in 2010.

Energy application Energy use

Hot water 118 TWh

Space heating 681 TWh

Process heat 536 TWh

Total heat use 1 335 TWh

Electricity use 582 TWh


7.  Future power supply systems  155

heat-recovery systems of cogeneration installations.


Hot water is needed in large quantities for such appli-
Hot water
cations as district heating systems, factories, hospi- Electrical
tals, hotels, and public swimming pools. In 2010, heat heating
element
demand exceeded electricity demand in Germany by
a factor of 2.3 (see table 7.2).
Cold water
According to Öko-Institut e.V, some 194 TWh or
14% of the total heat demand of 1335 TWh in Ger-
many was supplied by cogeneration installations in Grid
Burner Gas supply
2010. The bulk of heat demand is covered by burning
natural gas, heating oil, coal and wood. The estimated
installed electrical power of the cogeneration installa- Figure 7.13.  A hybrid hot-water boiler for
tions in Germany was 18 to 23 GW, or roughly 10% accommodating excess electricity.
of the total installed capacity of power stations. Tech-
nically, there are possibilities for having a substantially larger amount of cogeneration
installations. When there is excess electricity from wind turbines and solar panels,
the generators of the cogeneration plants can be switched off while the required heat
is produced by electric heating coils in the heat recovery boilers. This is far more
economic than investing in alternative storage techniques. The engine-generator

Geothermal
power Coal-based
until
Hydro obsolete
Bio waste
power
Bio waste 12 GW Nuclear-based
Wind
turbines until
obsolete Gasifier
8 GW
Solar PV
Gasifier panels 4 GW Agile
3 GW power
Gas-based generator
smart cogen
1 GW 4 GW Gas-turbine
Biomass
(co-)firing combined
cycle
1 GW
Pumped
hydro
Natural gas 250 GW Demand
resource & 3 GW management Natural gas
storage Heat use and system resource &
storage storage
1 GW
Export GRID
Import

Figure 7.14.  Suggestion for an optimum power supply portfolio for the 50Hertz TSO
case in 2012. The red arrows indicate energy flows towards and from the grid, while the
green arrows indicate fuel gas flows.
156  Power supply challenges

combinations of the cogeneration units will operate as dispatchable and flexible fuel-
based generators, should there be insufficient electricity from renewables. Denmark
is currently introducing this hybrid solution using electrical heating in combined
heat and power installations.
It might also be a good idea to start installing electrical heating coils to gas-fired
domestic heating systems and water boilers. This is not expensive. Installing electric
heat pumps in combination with a heat storage system is another promising option.
With further expansion of volatile renewable electricity sources, much more short-
term excess electricity will enter the supply system and that can be put to excellent
use to reduce the consumption of gas, oil, and coal for heating water.
Apparently, the German 50Hertz transmission system operator has been able to
keep the electricity system stable during 2012 with the help of existing power plants
and through exports and imports with neighbouring areas. A brand new dedicated
electricity supply system would, however, be different from the current one. A sug-
gestion for such a system is given in figure 7.14.
The maximum electricity demand in the region is 14 GW, as shown in the demand
distribution curve of figure 7.2. In addition to the installed 12 GW of wind-turbine
power and 8 GW of solar, 1 GW of biomass-based power is present in figure 7.14 as
a renewable resource. Furthermore, the 3 GW of agile cogeneration installations have
been equipped with electric heating coils in their heat-recovery systems. Also, 3 GW
of electricity can be absorbed by domestic and commercial boilers and heat pumps.
Pumped hydro and imports/exports are both given a capacity of 1 GW. An additional
250 MW of demand reduction has been made possible with smart appliances and
similar savings. This might mean temporarily switching off 125,000 washing machines
or tumble dryers. Apparently, maintaining grid balance with smart domestic appliances
requires quite an effort. A better solution might be to reduce the demand from large
users, such as cooled warehouses.
A total of 4 GW of combined cycle gas turbines has been added for covering the
intermediate load during times of low wind and sunshine. A further 4 GW based
on flexible and agile machinery has been added to provide power for fast ramping
up and down, to compensate for forecasting errors, and for non-spinning secondary
and tertiary reserves. The maximum size of an individual power plant in the system
should not exceed 300 MW. Otherwise, the tripping of a generator cannot be com-
pensated for properly by the primary reserves offered by the other generators.
The total capacity including the demand management system as shown in figure
7.14, is sufficient to cover maximum demand and to absorb the bulk of the excess
output from the renewable sources. The system can easily be prepared for a future
with double the amount of wind and solar-based generation by installing more elec-
tric heating using boilers and heat pumps. A fleet of hybrid and fully electric vehicles
will offer additional flexibility in electricity demand. The fuel-based power plants
will inevitably have a low utilisation factor, but their investment and operating costs
will be low due to the application of relatively cheap gas-based techniques and non-
spinning reserves.
7.  Future power supply systems  157

50 Hertz transmission, 50 Hertz transmission,


Germany, May 12, 2012 Germany, May 25, 2012

10 000

8 000
Power (MW)

Load
6 000
Others
4 000

2 000 Wind

0 Solar
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Time of the day (hours) Time of the day (hours)

Figure 7.15.  Power demand and supply in the 50Hertz TSO area for two different days
in May 2012.

7.4. Discussion regarding the suggested optimum power supply


portfolio
To give more detailed reasons for the suggested power supply portfolio in the previous
section, we use two more figures from chapter 3. The situation on May 12, 2012, as
depicted on the left side of figure 7.15, involves high output from wind turbines and a
moderate solar panel output. The red curve shows the electricity production that has to
be covered by other sources. The minimum in the red curve is a very low 1 GW. If we
exclude importing and exporting, this 1 GW has to be supplied by fuel-based power
plants. These plants also have to provide primary reserves and frequency control. In
order to guarantee system stability during malfunctions, the size of each power plant
should be restricted to a maximum of 100 MW. With these premises, the load of 1 GW
can be provided by 11 power plants, each running at 91 MW. Each of the 100 MW
power plants should be installed close to load centres and distributed throughout the
TSO area in order to avoid problems with the supply of reactive power.
Directly following the dip in the red curve ending at 6 pm on May 12, the
power plants have to quickly ramp up their output by about 1 GW per hour. This
is because the output from both wind and solar declined rapidly. Fortunately, this
happened at a moment when the total demand also decreased. The fast downward
slope in wind power output could also have happened 30 minutes earlier or later.
In the case of an earlier decline in wind output, the upward slope in the red curve
would have been much faster. Combined cycles require at least one hour to reach
full output, so they react too slowly for such an uncertain event. By contrast, flex-
ible power plants driven by fast-reacting engines can reach full output in less than
5 minutes. Hydropower might do the same job, but it is not abundantly available
in the area.
158  Power supply challenges

Electricity grid connection

1 2 3 4 5 6 7 8 9 10

Flexible fuel supply

Figure 7.16.  A flexible power plant based on multiple agile engines in parallel.

The suggested 4 GW of agile flexible generation in figure 7.14 can easily provide
the fast ramping required. If the weather forecast predicts that the output from the
renewable sources will remain low for eight hours or so, the suggested 4 GW of
combined cycles could be started up and a large number of the engine-driven units
could be gradually switched off. Some flexible power plants should stay online for
frequency control, load following, primary reserves, and for compensating for fore-
casting errors.
For satisfying the power demand on May 25, 2012, there is a sufficiently long
time span for the installed capacity of combined cycles to reach their full output of
4 GW. The peaks in demand of up to 8 GW can nicely be covered by agile power
generation. Even if more combined cycles were available, starting them up for these
peaks would not be worthwhile because
of the short time that they are needed.
Any residual demand can be covered 50
by the the projected cogeneration plant 48
Kilowatthour/capita/year

46
engines that would otherwise be utilised
44
little in May because of the low heat 42
demand. 40
Using batteries or other storage 38
devices might help a little for smoothing 36
34
the load patterns indicated in red in
32
figure 7.15, but the energy required to 30
charge the storage device has anyway to 2003 2004 2005 2006 2007 2008 2009 2010
be generated by fuel-based generators in Time (year)
this case. Investing in storage facilities
can only be considered if the renewable Figure 7.17.  Example of an emerging economy: the devel-
sources produce a substantial amount of opment of the per person electricity use per capita in Ethiopia
excess energy during the day for a large (data from IEA).
7.  Future power supply systems  159

fraction of the year. That might be the case if the renewable portfolio of the 50Hertz
TSO area will be double that of the situation in 2012. Even then, prolonged periods
of no wind and sunshine will occur, so backup power based on fuels is required
anyhow. It should be mentioned that agile flexible power plants do not require
smoothing of their output to achieve optimum efficiency.

When there is a high share of wind and solar power in the system, the fuel-based power
plants should be:

• Agile to provide faster ramping up and ramping down than has traditionally
been the case
• Fast so as to compensate for forecasting errors in the output from wind and
solar
• Not too big so that the tripping of a single plant can be compensated for by a
minimum amount of primary reserves
• Fast so that the amount of primary reserves can be limited, even with a low
amount of inertia in the system
• Fast to offer non-spinning secondary reserves
• Close to the load to avoid voltage collapse by the reactive power supply over
long transmission lines
• Inexpensive with low capital costs since their utilisation factor will be low.

7.5. An optimum electricity generation portfolio for emerging


economies
The World Bank estimates that some 1.3 billion people are still without electricity.
These people are primarily located in developing countries. To pull such countries out
of poverty and make them economically competitive with the rest of the world, their
per capita electricity use should reach at least some 1000 kWh/year by the year 2025.
This would drastically increase their productivity. Fortunately, many emerging econo-
mies currently show an annual growth in power demand of up to 10% per year. Figure
7.17 is an example from Ethiopia, with a steady rise in electricity use per capita per year.
Yet, this per capita electricity use is still extremely low compared to the world average
of 2500 kWh/year.
Developing countries generally have a poor infrastructure for supplying elec-
tricity. Many remote locations have to rely on small portable generators running
on expensive petrol or on batteries to provide a small amount of electricity. It is a
chicken and egg problem; the electricity demand is too small to make a proper distri-
bution system economic, while without electricity insufficient money can be earned
to build the required system. The best option appears to be to start in the same way
as Europe did a century ago. Small, municipal power plants can serve a local com-
munity, thus avoiding the costs of large transmission systems. Such power plants
should be based on multiple flexible units running in parallel to provide the required
160  Power supply challenges

reliability and efficiency. This solution also offers the possibility to expand the size of
the power plants gradually to meet growing demand. Rapid economic growth can be
expected as soon as a community gets access to reliable and affordable electricity.
The power plants should initially be able to run on different fuels. Many remote
locations do not yet have access to natural gas, so heavy fuel oil (HFO) is often the
only feasible energy source until such time that demand is high enough to allow the
construction of gas pipelines. Dual-fuel engines can easily run on both types of fuel.
Moreover, it is foreseen that bio fuels, wind turbines, geothermal power, hydropower,
and especially solar panels will play an increasing role in emerging economies. Sun-
shine is often abundant and the costs of solar panels are decreasing. Power plants
consisting of multiple agile units in parallel that can sustain many starts and stops
offer the proper backup for renewable sources. Most probably, emerging economies
will ultimately bypass the concept of using large conventional fuel-based power
plants.

Figure 7.18.  A typical village in an emerging economy

7.6. Conclusions
This chapter has discussed possible power plant portfolios for industrialised countries,
for countries with ambitious targets for renewable energy sources and for emerging
economies. It appears that in all cases a high share of agile, flexible and reliable power
plants is needed. Flexible power generation will be much more important than smart
appliances, smart meters, and smart grids. Flexible power plants are a good solution
for balancing electricity supply systems and are needed to bridge periods of low output
from renewables. They are essential for developing affordable, reliable and sustainable
power supply systems.
8 Power supply
challenges – A review
The previous chapters describe the challenges of maintaining a stable , reliable and
affordable electricity supply with much intermittent renewable power in the system.
Reduced rotating inertia requires power plants with high agility. Large fractions of
reactive power cannot be supplied via transmission lines. A disappearing baseload
demands more flexible power plants that are capable of starting and stopping
frequently. The utilisation factors of fuel-based power plants will decrease.
The crucial decisions required for obtaining a more sustainable global energy
supply have to be based on a thorough knowledge of all the aspects described
in this book. In any case, future power plants have to be agile and flexible. This
chapter reviews the issues so as to remind the reader that optimum solutions
depend to a large extent on variable boundary conditions.
164  Power supply challenges

8.1. Realism is needed in the energy debate


Energy is the driving force behind productivity, product quality, and comfortable living.
Energy is, therefore, the engine of the economy. Energy will be supplied to end-users
increasingly as electrical energy, because electricity is so versatile and convenient. Ulti-
mately, the world’s energy supply has to be sustainable, with acceptable short- and long-
term impacts on the environment. Yet, at the same time, a sustainable energy supply has
to be affordable and very reliable.
In 2011, the share of electricity in the world’s total primary Other
Oil 4.5
energy supply (TPES) was only 12%. Despite continuous 4.8
improvements in power plant efficiency, still a considerable 34%
of global TPES was needed for electricity production. Almost
Nuclear
70% of the primary energy required for electricity production 11.7 Coal
41.3
stemmed from coal, natural gas and oil. Hydropower and nuclear
power were responsible for the bulk of the remainder, while Hydro
15.8
electricity sources based on solar radiation, wind and biomass
produced less than 4.5% of the electricity demand. Natural gas
21.9
To replace the 3 150 Mtoe (megatonne oil equivalent) of
fossil fuels needed for electricity production in 2011 by renewable
energy sources is a huge task. A full withdrawal from the use of Figure 8.1.  Energy sources for
fossil fuels might not be possible for many decades to come, espe- electricity production in 2011 (data
cially since the demand for electricity in the world is expected to from IEA Key World Energy Statis-
tics 2013).
double over the next 20 years. The world’s population will grow
and electricity demand in emerging economies will increase
rapidly. In industrialized countries, electric heat pumps will often replace traditional
heating systems in homes and electric vehicles will be common on the roads. The
world economy will, therefore, increasingly depend on electricity – and much more
of it will be needed than now.
Policy makers and citizens cannot be blamed for lacking a clear view of the
electricity supply challenges ahead. Newspapers and the internet bring exciting news
about wind farms and solar installations that would produce enough power for, say,
100000 households. The fact that households consume on average only 20% of the
electricity needs in a modern nation is not explained. Globally, industry uses some
46% of the electric energy supplied. The remaining 34% is for services. Moreover,
the unavoidable need for backup generating capacity for renewables is hardly ever
discussed in the news.
In many areas in the world, the capacity factor of a wind farm is lower than 25%,
meaning that on average the output is less than 25% of the installed power capacity.
Solar panels often have a lower capacity factor than wind turbines; in Germany
for example it averages 9.5%. This is partly caused by the natural absence of solar
radiation during the night, but also by clouds. In very sunny places, as in the Sahara,
solar panels might reach a capacity factor of 30%. We read news about solar panels
reaching grid parity, meaning that the costs of electricity from a solar panel can
8.  Power supply challenges – A review  165

Capital costs solar panels; investment 1 716 €/kW


50

DR = 2.5%; life = 20 years


Capital costs (€cnt/kWh)

40
DR = 2.5%; life = 40 years
DR = 10 %; life = 20 years
30
DR = 10 %; life = 40 years

20

10

0
5 10 15 20 25 30 35 40

Capacity factor (%)

Figure 8.2.  Capital costs per kWh produced for solar panels depend on the boundary
conditions (DR = discount rate).

compete with the electricity costs from the grid. However, such statements can be
made only if the boundary conditions are known. First of all, the costs of installation
and grid connections have to be included. Figure 8.2 gives an example of the capital
costs of solar panels per kWh produced, depending on panel life and the discount
rate DR (the interest rate) that the owner requires.
If a private consumer is happy with a discount rate of 2.5 % and the panel life is
20 years, the costs of electricity from PV panels in a sunny location might approach
5 €cnt/kWh, which easily results in grid parity. For a professional investor who
desires a 10% discount rate in an area where the capacity factor of PV would only be
10%, electricity from an identical panel costs almost 25 €cnt/kWh. Panel life does
not have a large effect on the capital costs when the discount rate is high. Conversely,
with a low discount rate the effect is considerable. In conclusion, to talk about the
grid parity of renewable energy sources without giving the boundary conditions is
misleading.
When the costs of renewable energy are discussed, the costs of backup capacity,
either from storage systems or generators, are generally not taken into account. Yet
backup is always needed for solar panels and wind turbines. Chapter 6 has shown
that the actual price a consumer pays for electrical energy can be higher by a factor
of three than the production costs. This is because of profit rates, distribution costs,
government levies and taxes. Therefore, the real cost of creating a sustainable elec-
tricity supply depends on many factors.

8.2. Low capacity factors escalate balancing issues


One concern with renewable electricity sources is the effect of priority feed-in, and even
unconstrained feed-in, on supply system stability. If a country wants to cover one third
166  Power supply challenges

Maximum Average Minimum


175

150
% of average power demand

125

100

75

50

25

0
Power demand Wind 20% cap factor Wind 35% cap factor

Figure 8.3.  Low capacity factors of renewable energy sources result in high variability
in their output.

of its average electricity demand by wind power, the disturbing effect depends heavily
on the capacity factor of the wind turbines. This is illustrated in Figure 8.3.
If the capacity factor for wind parks is 20%, which is normal in a country like
Germany, and a third of the electricity demand has to be covered by wind, the
maximum wind output can exceed even the maximum demand for electricity. As a
result, the grid is heavily disturbed and curtailment of wind output is often required.
However, if the capacity factor is 35%, as is possible in optimum locations, the max-
imum in wind output is slightly less than the average demand for electricity. In this
case, the wind output exceeds demand only occasionally during times when demand
is below average. Figure 8.3. illustrates this. Identical situations can occur with solar
power.
Another effect of having a substantial share of electricity from renewables is that
the remaining fuel-based generators often have to very quickly ramp their output
up and down. They have to start and stop frequently, and have to compensate for
substantial forecasting errors. At the same time, these power plants need to be much
smaller in capacity than in the past in order to ensure grid stability during contin-
gencies.
Solar panels in hot climates, where peak demand for electricity coincides with
peak sunshine, can have an effective smoothing impact on what the fuel-based gen-
erators have to supply during the daytime. However, where a large fraction of elec-
tricity demand is covered by solar panels, other generating capacity will experience
high ramping up rates in the evening when the sun sets.
8.  Power supply challenges – A review  167

8.3. Flexible local generators of limited size offer excellent backup


for renewables
It has been suggested that the task of balancing renewable energy could be handled by
transporting excess electricity from one country to another. Current transmission lines
were not intended for this purpose. The investment costs of such a system would be very
high while its utilisation factor would be low. Moreover, the energy loss during transport
would be high.
The situation is worse in the case where renewable sources do not produce a pro-
portional fraction of reactive power. Long overhead transmission lines between gen-
eration and demand lose their capacity should they have to transport larger fractions
of reactive power. This results in a local collapse of voltage, i.e. a black out, especially
when active power demand is low.
Moreover, high winds generally cover a substantial part of continents so that
peaks in wind turbine output in adjacent countries generally coincide. Building
extensive transmission systems for transporting electricity will therefore not help
in solving the problems arising from intermittency of renewable energy sources.
In addition, large distant power plants offer insufficient reserve capacity in case of
calamities. Multiple smaller local generators are needed instead – in fact they are the
solution for all problems mentioned.

8.4 Natural gas is ideal for backup capacity


The media and some researchers often give the impression that there are high hopes for
developing electricity storage systems to smooth demand fluctuations and even out the
varying output of renewable sources. This appears to be possible only in locations with
guaranteed daily sunshine or with winds without long doldrums –  meaning the gaps in
output are less than 24 hours. Then batteries, pumped hydro and compressed air might
have sufficient capacity. Nevertheless, such storage adds at least 10 €cnt per kWh to the
cost of electricity.
Covering a windless time span of 10 days from storage will be prohibitively
costly, let alone storing solar energy from the summertime to be consumed in the
winter. Demand-side response management helps only for short-term balancing.
Moreover, it is extremely difficult to find sufficient switchable load capacity. To create
a demand response of 1 GW with smart appliances in households, at least 500 000
active washing machines or laundry dryers of 2 kW each have to be controlled. If
their utilisation factor is 5%, ten million smart appliances are needed for creating the
1 GW demand response. The practical realisation of demand response from house-
holds seems to be wishful thinking, especially since households consume on average
only 20% of all electricity produced.
Only a large-scale hydropower system is capable of providing long-term storage.
Norway is an example. The so-called power-to-gas concept of producing hydrogen
with excess electricity and injecting it into natural-gas pipelines for later use, has an
168  Power supply challenges

Figure 8.4.  Electricity is all around us – even when we are unplugged.

energy efficiency as low as 25% and would add to the costs of electricity consider-
ably. Moreover, hydrogen deteriorates the quality of natural gas.
Natural gas in combination with smart agile generators offers an excellent option
as backup systems for keeping electricity systems balanced over extended time spans.
Using natural gas only as a transient fuel and burning it all as quickly as possible
during the first half of the 21st century is not wise. One should save the gas as much
as possible for creating flexible backup over a much longer time span.

8.5. Integrating power demand and heat demand offers good


perspectives
A very effective way to avoid grid instability caused by excess electricity from
wind turbines and solar panels is an integrated approach with heat demand. Demand
for heat in the world is still much higher than the demand for electricity. Excess
electricity from renewables can easily be fed into hot-water boilers and storage tanks
with electrical heating coils. District heating systems are an interesting application.
Where there is insufficient electricity production from renewable sources, combined
heat and power (CHP) units can provide the needed electricity and heat. In the case
of excess electricity, the cogeneration units can switch the engines off while the water
is heated by electricity. The engine-driven gas-fuelled generating units of cogenera-
tion systems can also provide the necessary ancillary services, such as compensating
for forecasting errors and providing reserve capacity.
8.  Power supply challenges – A review  169

8.6. Agile, flexible power plants help to ensure a reliable and cost-
effective power supply
Ideally, flexible power generation is based on local power plants each having multiple
identical generating units in parallel. The size of each individual generator in such power
plants is limited so as to create sufficient flexibility in output and to guarantee a high
combined availability. The investment costs in such a system are relative low because
of the uniformity and series production, while the lead time needed for engineering,
procurement and construction is short because of the standardised units. Maintenance
costs per kWh and fuel efficiency are independent of the power plant’s output. The gen-
erators can be switched on and off without reducing the interval between maintenance
work. The generators provide non-spinning reserves, as well as adequate reserves for
forecasting errors since they can start up in less than a minute and deliver full output
within 5 minutes. Maintenance is easy since it is fully standardised.
The agility of these power generators makes them suitable for lifting the burden
of frequency control and rapid up and down ramping from less flexible power plants.
They can provide fast primary reserves that are needed when inertia in the system is
diminishing. They can also offer secondary reserves without spinning. These advan-
tages will lower the total costs of electricity production.
In systems with a substantial amount of renewable electricity sources, smart fast-
reacting power generators offer the optimum solution to keep the system balanced

Figure 8.5.  Power systems can be reliable, affordable and sustainable in the future – if
we make the right choices now.
170  Power supply challenges

in an affordable and reliable way. Furthermore, smart poweragile generators are the
ideal solution in emerging economies, where power demand is initially low but likely
to accelerate rapidly. Their output flexibility and load-independent fuel efficiency and
maintenance costs make them the cheapest option for guaranteeing high reliability
in the electricity supply.

8.7. Outlook for the future


The goal of attaining a sustainable, reliable, and affordable electricity supply involves
a huge challenge. Reducing CO2 emissions with either renewable energy sources or
carbon capture and storage (CCS) will easily double electricity production costs. In a
closed economy this would not create problems since any expenses for the equipment
and for the resources will stay within that economy. It will even create jobs. The intrinsic
value of electricity is so high that a doubling of electricity costs would still leave them
well below the true value of having electricity. In reality however, economies are open
and energy intensive production will move to areas with the lowest electricity costs if
governments do not create special incentives for the industry.
Yet, even if a doubling of electricity production costs is very likely in a more
sustainable world, accepting an even larger increase in costs should be avoided. In
a desperate attempt to quickly reach preset goals for lower emissions and having a
lower dependence on fuel imports, decision makers might opt for economically unat-
tractive solutions.
This book has shown that flexible and agile local power plants of a limited
size can excellently serve as a facilitator for integrating renewable energy and for
improving the fuel efficiency of less flexible power plants. Flexible and modular
power plants are also the perfect solution for providing the world’s electricity-
deprived population with this resource to help them in creating wealth. Natural gas
is an ideal backup fuel and should be saved for that purpose as much as possible. Gas
companies have to ensure a good quality of gas for electricity generators, since the
power sector will be their largest customer in the future.
172  Appendix  1

Appendix 1
Background physics and mathematics regarding the role of inertia in power systems

A car with a mass m at standstill at the top of a slope has so-called potential energy due
to gravity. Gravity is the force with which the earth ‘pulls’ at the car. This force is pro-
portional to the mass m of the car and the factor g. The factor g is called the acceleration
of gravity. The value of g depends on the position of an object on earth, but generally
a value of 9.81 m/s2 is used. The potential energy Ep of the car at the top of the slope
with a height h equals:
Ep = m · g · h Equation A1.0.

As soon as the brakes of the car are released, the static force of gravity starts the car
moving down the slope. Without an external force, the mass of the car would never
move. The mass of an object can be seen as resistance against change in motion: mass
means inertia. Because of the force of gravity, the car develops speed. This speed is called
velocity v in physics. If road friction and air resistance are neglected, all potential energy
will be converted into speed-related energy, known as kinetic energy in physics. The
kinetic energy Ek of an object depends on its mass and its speed:
E k = ½ mv 2 Equation A1.1.

If the height of the slope is 50 metres and the mass of the car is 1200 kg, the potential
energy Ep of the car equals:

Ep = 1200 ∙ 9.81 ∙ 50 = 588600 kg m/s2 m = 588600 Nm = 588600 J ≈ 0.59 MJ,

since 1 N (newton) = 1 kg m/s2 and 1 J (joule) = 1 N m.

The international system for units uses the following prefixes to shorten large numbers:

k (kilo) = 1000
M (mega) = 1000,000
G (giga) = 1000,000,000
T (tera) = 1000,000,000,000
P (peta) = 1000,000,000,000,000
E (exa) = 1000,000,000,000,000,000
Z (zetta) = 1000,000,000,000,000,000,000
Y (yotta) = 1000,000,000,000,000,000,000,000
Appendix  1  173

As soon as the car has reached the bottom of the slope, all its potential energy has been
converted into kinetic energy. Equation A1.2 reveals that the car will have a speed of:

v = √ (588600 ∙ 2/1200) = 31.3 m/s, equalling 112.8 km/h.

If the brakes of the car are then applied, the kinetic energy will be converted into heat.
The 0.59 MJ of kinetic energy is enough to heat up water for only 13 cups of tea. This
illustrates that storing large amounts of energy as potential energy via, for instance, ele-
vating masses of pumped hydro, or as kinetic energy in a flywheel, is not an easy task.
A physical object that can rotate around a shaft also has inertia. It means that
without an external force, or momentum in this case, the object will not start to
rotate. Conversely, if the object is rotating, it will need an external force to slow it
down again. The momentum applied delivers energy to the object, and that energy is
again stored as kinetic energy in the object. The inertia of objects that can rotate is
called the moment of inertia I.
The moment of inertia of an object depends on its mass m and the distance r of
that mass to its shaft. A disk has a moment of inertia I equal to ½ m r2 , and the I of a
thin ring equals m r 2 .

A disk: Thin ring:

The moment of inertia of a generator together with the prime mover that supplies the
generator with energy closely resembles that of a disk.
The kinetic energy Ek of a rotating object equals:

Ek = ½ I (2 π f)2 = 2 π2 I f 2 Equation A1.3.

in which f is the number of revolutions per second in Hz (hertz).

For a small change in frequency, the derivative of equation A1.3 gives the amount of
change in the rotating energy:
dE
= 4π 2 I f Equation A1.4.
df
174  Appendix  1

Or:
dE = 4π 2 I f df Equation A1.5.

For a change of the rotating energy in time, we write:


dE df Equation A1.6.
= 4π 2 I f
dt dt
The change in rotating energy of a generator is the difference ΔP in power supply to
the generator by the prime mover and the power demand from the connectors of the
generator.

Therefore, we can write:


df Equation A1.7.
ΔP = 4π 2 I f
dt
and:
df ΔP
= 2 Equation A1.8.
dt 4π I f
For power systems, the inertia of a generator is generally expressed in the inertia con-
stant τi.
2π 2 I fn2
τi = Equation A1.9.
Pn
in which Pn = the nominal power of the generator and fn the nominal frequency of the
generator.

By combining the two last equations, for the instantaneous change in frequency due to
unbalance in power we can write:
df ΔP 2 1
= If Equation A1.10.
dt Pn n 2fτi
At t = 0, when the unbalance ΔP just occurs, f = fn, so:
df ΔP fn
at t = 0 = Equation A1.11.
dt Pn 2τi
This means that the initial inclination in speed change for a system with an inertia constant
τi is determined by the fraction of the unbalance in power ΔP of the nominal power Pn.

Equation A1.10 can be rewritten as:


ΔP fn
2
f df = dt Equation A1.12.
Pn 2τi
Appendix  1  175

By integrating this equation, we get:


ΔP fn
2
f2 = t+C Equation A1.13.
Pn τi
At t = 0, f 2 = (fn)2, so that C = (fn)2.

Therefore, the change in rotational frequency f versus time equals:


ΔP fn2
f = √(f n2 + t) Equation A1.14.
Pn τi
Equation A1.14 applies for a constant unbalance between the power supply and the
power demand, meaning that the load is not affected by the frequency.
In reality, the self regulating power in the grid will ensure that the unbalance is reduced
when the frequency changes.

In that case, equation A1.10 has to be rewritten as:


a
df ΔP – 100 Pinitial (fn – f) 2 1
= fn Equation A1.15.
dt Pn 2fτi

In which a is the percentage of load change per Hz and Pinitial is the initial load.
As soon as the primary control reserves Ppc step in, the unbalance will be further reduced
and equation A1.15 has to be rewritten as:
a
df ΔP – 100 Pinitial (fn – f) + Ppc ) 2 1 Equation A1.16.
= fn
dt Pn 2fτi

Equations A1.15 and A1.16 are quite complicated to integrate mathematically, but
numerical integration with small steps in time gives a good approximation.
Figures 2.7, 2.11, 2.12 and 2.13 in Chapter 2 are the results of the numerical integration
of equations A1.15 and A1.16.
176  Appendix  2

Appendix 2
Power plant performance during a major fault in the system

Faults and failures are unavoidable and occur in any system. Faults in electricity supply
systems generally mean deviations from a normal situation caused by external events,
such as a falling tree, lightning, animal actions and vandalism. So-called failures result
from equipment malfunctioning and weak grids. Human error is also a cause of unde-
sired occurrences. A failing generator resulting in a power plant trip causes a sudden
unbalance between electricity production and consumption, as described in chapter 2.
For a generator, a short circuit in a transmission line is a fault to which it will respond.
This section will describe in general terms what happens with a generator in the
case of a short circuit in the grid. A detailed technical description of all the dynamic
events would, however, require a separate book. Here, it is important to understand
what the consequences of a short circuit are for a power plant and the grid. Fortu-
nately, grids are equipped with instruments that detect anomalies. As soon as a
major problem such as a short circuit is detected, the relevant grid section will be
isolated via breakers. Modern grid protection equipment is capable of taking such an
action within 150 milliseconds. Nevertheless, during a limited time, a short circuit
can have a severe effect on grid voltage, leading to a substantial loss of load for gen-
erators that sense the voltage.
With a so-called bolted short circuit, the voltage drops to zero and the generator
loses its external load. A stepwise loss of load implies that a generator will accelerate

Fuel Speed/load
supply controller

n
P

Prime mover Coupling GENERATOR


V = C.N.Ф
V,I

Flywheel

Voltage/power
factor
controller

Figure A2.1.  Control circuit of a prime mover driving a generator in a generating set.
The prime mover cannot instantaneously reduce its power output should there be a loss
of load caused by a short circuit in the grid.
Appendix  2  177

100
90
80
70
Voltage (%)

60
50
Energinet dk
40 German Transmission Grid Code
(FRT to zero voltage depent on distance of the WT from PCC)
30 National Grid
20 PSE-Operator
REE
10
0
–500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Duration of short circuit (ms)

Figure A2.2.  Examples of different fault-ride through rules in Europe prior to a uniform
ENTSO-E rule being implemented.

its rotational velocity for a while. The reason is that the prime mover that is driving
the generator cannot reduce its power output immediately. If the generator has a
higher speed than the grid frequency, the voltage of the generator can be consider-
ably turned out of phase with respect to the voltage of the main grid. In that case,
a return of the grid voltage after the corrective action by the grid protection system
can cause damage to the generator set and lead to system instability.
The risks of damage to the generating set and instability of the grid are the rea-
sons that in the past local generators, such as cogeneration installations and wind
turbines were disconnected from the grid as soon as the voltage exceeded specified
limits. Reconnection took place after the grid voltage returned to normal. How-
ever, with a substantial amount of decentralised generators in a system, this tradi-
tional approach might result in the available generating capacity immediately after
clearance of the fault being insufficient. The relative number of large, high inertia
generating units that traditionally helped to run throughout grid faults can be low
when output from wind turbines and solar PV panels is significant. This is why grid
operators nowadays specify that all generators supplying energy to the grid should
have fault-ride-through capability, so that they can support the grid during and after
a calamity. It means also that the generator has to provide reactive power to the grid
during the occurrence. Afterwards, the generator has to deliver the same power to
the grid as before.
As mentioned earlier, if the load of a generator instantaneously disappears, the
fuel supply to the prime mover that drives the generator cannot immediately be
reduced. For turbines, it takes time to decrease the opening of the fuel supply valves
178  Appendix  2

or steam valves. Too drastic reductions Induction coils Rotor


in the fuel supply can cause flame out
in the combustors of gas turbines. For a
modern reciprocating engine, the typical S
four-stroke process means that almost Slip rings and
contact brushes

N
two revolutions of the engine shaft are
Direct current for

N
needed to consume the fuel already sup- S the magnetic field
plied to the cylinders. Two revolutions
of an engine running at 750 rpm already Drive shaft AC voltage
occupy 160 ms. The fuel supply to a
prime mover should not be drastically
Electric energy
cut off during a short circuit, since after
the occurrence the rules declare that
the generating units should again deliver Figure A2.3.  Schematic representation of a generator with
their initial output. two pole pairs: m = 2. (picture taken from google images).
In a first approach to study the effect
of a short circuit on a generator, a synchronous generating unit can be seen as a
prime mover, such as a turbine or an engine, which drives a rotating magnet in a set
of stationary coils. The rotating magnet is called the rotor and the stationary part is
called the stator. The voltage V generated in the stationary coils is directly propor-
tional to the running speed n and the magnetic flux Φ from the rotor:

V= c·n·Φ Equation A2.1

In grid-parallel operation, the grid dominates the frequency and voltage. For the
generator to produce electric energy, the angle of the rotor’s magnetic field had to
lead the phase angle of the voltage of the stator by the so-called load angle δ. Simply
said, the magnet has to pull at the grid via a virtual spring in order to transfer energy
to the stationary coils. The load angle δ increases with the relative power output of
the engine-generator combination. Under steady-state conditions, the load-angle
curve versus the generator output has a roughly sine-wave shape: if the load angle
exceeds 90°, the generator has reached its pull-out or tipping torque and the system
becomes unstable. The load angle also depends on the strength of the magnetic field,
where a stronger field (= more magnetisation) leads to a smaller load angle. Simply
said again, if the pulling spring is stiffer, the spring expands less for a given force.
For commercial generators, the electrical load angle ranges between 15° and 60°.
The physical, or mechanical, rotor angle δlm can be found by dividing the electrical
load angle δle by the number of pole pairs m. This pole-pair number m is 1 for an n
= 3000/min machine in a 50 Hz grid, and m is 4 for a generator running at 750/min
in a 50 Hz grid. With a stronger magnetic field from the rotor than is necessary for
generating the required voltage, the generator current will lead the voltage (capaci-
tive character). This means that the generator produces the reactive current needed
for the inductive loads of the grid. In this case, the load angle is smaller than in
Appendix  2  179

3.0
Pull-out torque or
tipping torque
2.5

2.0
Power P

1.5

1.0

0.5

0.0
0 20 40 60 80 100 120 140 160 180
Electrical load angle (deg)

Figure A2.4.  Example of steady-state generator output versus electrical load angle δle
for a typical electrical load angle of 24° and a nominal power P of 1.

the case of a rotor field that is just sufficient to produce the necessary voltage. With
lower excitation than needed to produce the grid voltage, so-called under-excitation,
the generator needs reactive current from the grid and the load angle is larger. With
a larger load angle, the stability of the generator is smaller. The ‘spring’ is weaker in
this case. If the load angle exceeds 90°, the so-called pull-out or tipping torque is
passed and the generator will start to spin faster than the grid. This is called pole
slip.
With a synchronous generator unit in grid-parallel operation, two extreme situa-
tions can occur: instantaneous disconnection from the grid by the generator circuit
breakers, or a short circuit in the connected grid. In both cases, the load of the gen-
erator is lost. Loss of the electric load means that the driving energy from the prime
mover will start to accelerate the generator.
A simplified example will now show how the speed of the generator and the
angle of its poles with respect to that of the grid might change in the case of a full
loss of load. In reality, the situation with a short circuit is not so dramatic, since the
grid voltage generally does not drop to dead zero. We can re-write equation A2.3 in
terms of the angular rotor speed ω = 2πf, and describe how the running speed of the
generating unit will increase in time (= accelerate) with a loss of power ΔP:
dω πΔP
= f Equation A2.6.
dt τi Pnominal nominal
180  Appendix  2

The actual angular speed increase Δω can be found by integrating equation


A2.6:
πΔP
Δω = f t Equation A2.7.
τi Pnominal nominal
The mechanical load angle increase Δφlm is found by integrating equation A2.7
and equals in radians:

Δφlm = πΔP
f t2 Equation A2.8.
2τi Pnominal nominal
Equation A2.8 can be re-written in degrees instead of in radians (π = 180°). For a
full loss of load, ΔP equals Pnominal, so that:

Δφlm = 180 fnominal t 2 Equation A2.9.


2τi
Figure A2.5 illustrates the simplified case of how the running speed of a
generating set with 4 pole pairs will increase if all of the full load is lost while the
prime mover continues feeding energy to the generator as needed for full load. In
this example, an inertia constant τi of 1.5 s has been chosen. The figure also shows
how the mechanical load angle δlm and the electrical load angle δle change in time
with respect to the central grid of a constant frequency of 50 Hz. It will be clear

100 790
Mech load angle
Electr load angle
Mechanical + electricial load angle (deg)

Speed (rpm)
80 780
Speed (rpm)

60 770

40 760

20 750

0 740
0 25 50 75 100 125 150

Time (ms)

Figure A2.5.  Example of a change in running speed and load angles with respect to the
central grid in the simplified case of a fully lost generator load.
Appendix  2  181

that recurrence of the full grid voltage after 150 ms will show a large mismatch in
running speed, voltage and phase angle. The situation when the grid returns can be
compared with that of connecting a generator to the grid with a mismatch in syn-
chronisation.
In reality, the case is much more complex. Between the location of a short cir-
cuit in the grid and the generator, generally a line section and a transformer with
impedance are present. In addition, a generator is equipped with damping windings
and has internal impedance. This will mitigate the extreme reaction as given in our
example. The real-life situation is so complex that only computer modelling with
many parameters can approach what will actually happen in practice.
In the context of this book, it is important to understand that a rapid temporary
reduction in power supply from the prime mover to the generator during a grid fault
helps to avoid a severe mismatch when the fault has been cleared. A flexible power
generator is able to sense the drop in voltage caused by the short circuit, and to react
with a temporary instantaneous decrease in output. Combustion engines have the
ability to retard the onset of combustion in the cylinders instantaneously, resulting in
a direct decrease in power output. When the grid voltage returns, the initial timing
will be immediately restored. Such a rapidly reacting facility creates the flexibility to
ride through a grid fault.
Dr. Jacob Klimstra
Senior Energy Specialist
‘It Hazzewâld’
Hazzeloane 3
Broeksterwâld
The Netherlands

Jacob Klimstra studied electrical, electronic and mechanical engineering. He worked


29 years at the research laboratory of N.V. Nederlandse Gasunie on pulsating combus-
tion, vibration-based machinery diagnostics and reciprocating machines. From 1993 to
early 2000, Jacob was Head of Department of Industrial Gas Applications at Gasunie
Research. From 2000 until the end of 2009, Jacob was employed by Wärtsilä as a senior
specialist for engine-driven power systems. From 2010, Jacob serves the industry as a
consultant.
Jacob received the Richard Way Memorial Prize for his Ph.D. thesis, the Van
Oostrom Meyjes Prize from The Royal Netherlands Institution of Gas Engineers and
5 Oral Presentation Awards a well as the Distinguished Speaker Award from SAE. In
September 2000, he received the ICE Division Speaker Award from the American
Society of Mechanical Engineers. In 2010, he received a lifetime recognition award
from Cogen Europe.
He is the co-author of the book Smart Power Generation published in 2011.
Jacob also teaches at short courses on energy and engine technology at universities
around the world. In addition, he serves at the Advisory Board of PowerGen confer-
ences and acts as moderator and speaker at many international conferences. Since
the beginning of 2014, he is managing editor of the Cogeneration and On Site Power
Production magazine.
References
Books, papers, reports and magazines:
Vuorinen, Asko, ‘Planning of Optimal Power Systems, 2009 edition, ISBN 978-952-
67057-1-2, Ekoenergo Oy, Espoo, Finland, 2009
Klimstra, Jacob and Markus Hotakainen, ‘Smart Power Generation’, ISBN 978-951-692-
846-6, Avain Publishers, Helsinki, Finland, 2011
Dekker, G. and J. Frunt, KEMA report 74100846-ETD/SDA 12-00079, Arnhem, The
Netherlands, March 16, 2012
International Energy Agency, ‘World Energy Outlook 2012’, ISBN 978-92-64-18084-0,
Paris, France, 2012
International Energy Agency, ‘Key World Power Statistics’ (years 2000 – 2013)
The Economist, vol. 407, nr. 8830, April 2013
Chabot, Bernard, ‘Renewable Energy for Electricity in California in 2012 and its Future
Role’, http://www.renewablesinternational.net/
Eurelectric, ‘Hydropower for a sustainable Europe’, Eurelectric Fact Sheets, February
2013
Appleyard, David, ‘Tracking the Price of U.S. Grid-connected PV’ Renewable Energy
World Conference, Mumbai, India, 6-8 May 2013
ENTSO-E, ‘ Network Code for Requirements for Grid Connection Applicable to all
Generators, Brussels, Belgium, 8 March 2013
Martens, Deborah, editor, ‘ Joint EASE/EERA recommendations for a
European Energy Storage Technology Development Roadmap towards 2030’ European
Association for Storage of Energy, March 2013
Seamus Garvey, ‘Compressed Air Energy Storage (CAES): Cycle Efficiency’,
Pre-Conference Workshop on Energy Storage, Marcus Evans, Amsterdam, 1st December
2010
Jacob P. Aho, Andrew D. Buckspan, Fiona M. Dunne and Lucy Y. Pao, ‘Controlling
Wind Energy for Utility Grid Reliability’, ASME Magazine MECHANICAL ENGI-
NEERING, September 2013
Websites:
www.eirgrid.com/operations/systemperformancedata/
www.sepco-solarlighting.com/blog/bid/102981/Renewable-Energy-and-The-Indus-
trial-Revolution-Part-1
http://content.caiso.com/green/renewrpt/20130610_DailyRenewablesWatch.txt
http://www.swissgrid.ch/swissgrid/de/home/reliability.html
http://www.tennet.org/bedrijfsvoering/xmldownloads/index.aspx
http://www.amprion.net/en/grid-data
http://www.50hertz.com/cps/rde/xchg/trm_de/hs.xsl/Netzkennzahlen.html
https://www.entsoe.eu/data/data-portal/
http://www.ree.es/en/activities/realtime-demand-and-generation
http://www.power-thru.com/
http://www.met.ie/
http://en.ilmatieteenlaitos.fi/
http://www.we-at-sea.org/een-energie-eiland-voor-de-kust-van-nederland/
http://www.ercot.com/gridinfo/
http://www.marketsandmarkets.com
www.europe.eu
www.eia.gov
www.eurelectric.org
www.ercot.com
www.caiso.com
Glossary
Active power – Product of voltage and maximum load, thus ensuring high fuel
current resulting in a net release of efficiency and low maintenance costs.
electrical energy. Cogeneration – Effective method for util-
Ancillary services – Services provided by ising the heat released during the pro-
electricity generators to transmission duction of electrical energy for process
system operators (TSOs) other than heating, space heating or cooling.
producing electrical energy directly, Combined cycle – Process where the heat
such as frequency control, backup released by a gas engine or gas turbine
capacity, and spinning and non-spin- is recovered and turned into steam for
ning reserve. a steam turbine driving an additional
Availability – Condition in which a generator.
machine is ready to perform the duty Compressed air energy storage – System
for which it is intended. that uses electrical energy to drive an air
Balancing – Controlling electricity pro- compressor to store compressed air in
duction so that it fully matches elec- a cavity for later use with an expander-
tricity demand. generator combination during times
Baseload – Constant demand level for of peak electricity demand, or for bal-
electrical energy that is present during ancing electricity supply and demand.
a prolonged period of time. Compression heat pump – Machine that
uses a motor-driven compression and
Black start capability – Capability of a expansion process to bring heat from
generating unit to start up without sup- one temperature level to another, for
port from an external electricity source, heating or for cooling.
such as the electricity grid. Contingency – Abnormal condition or
Calorific value – Energy content of a fuel situation leading to the close to step-
expressed per unit of volume or unit of wise increasing or decreasing of gener-
mass. ating power.
Capacity factor – Ratio of actual elec- Crest factor – Ratio of the peak value of a
trical energy that a generating set can variable quantity and its average value.
produce in a certain time span and Demand side management – Method
the amount of electrical energy that it for decreasing electricity demand by
could produce if running at full output switching off part of the electricity con-
during that time span. sumption.
Carbon capture and storage – Process of Depth of discharge – Ratio of the amount
capturing the carbon dioxide emissions of energy that can be taken from an
from the use of fuels and storing them energy storage device and the stored
indefinitely, for example in geological energy in that device, without dam-
formations. aging the storage device.
Cascading – Method for an array of elec- Discount rate – Fraction of an invested
tricity generators in parallel to run capital that is desired as an annual
individual generators only close to yield.
Dispatchability – Capacity of a generating turbines and the actual required
unit to deliver a certain performance as output.
required by the generator operator or Frequency – Number of repetitive cycles
transmission system operator. per second of a process, with the unit
Distribution grid – System that distrib- Hz (hertz).
utes electricity or gas to households, Gas engine – Machine that converts
commercial users, and small indus- chemical energy stored in fuel gas into
tries. mechanical energy.
Droop – Dependence of the output from Gas turbine combined cycle – Combina-
a generator on a deviation from a fre- tion of a gas or oil-fired turbine and a
quency setpoint steam turbine. The steam turbine uses
Electricity demand pattern – Hourly, heat from the exhaust gas of the gas
daily, weekly, monthly, or annual pat- turbine.
tern of electricity use, including base- Generating portfolio – Combination of
load, intermediate load and peak load. all the power plants in a given area, such
Electricity highway – Buzzword for a as nuclear plants, coal-fired plants, gas-
high-capacity system for transporting fired plants, hydropower and renewable
electrical energy over longer distances. electricity generators.
Electricity intensity – Amount of elec- Gross domestic product (GDP) – Total
trical energy needed to create a certain monetary value of the amount of goods
amount of gross domestic product, and services produced per year in a
often expressed in kWh/€ or kWh/$. country. Often, the GDP is expressed
Energy – Amount of physical work stored in the local purchasing power parity
or delivered to a process. (PPP) of the US$, since the buying
Energy storage – Storing energy for later power of the US$ differs from country
use, often in pumped hydro, batteries, to country.
flywheels, and compressed air, but pri- High-voltage ac – Three-wire system for
marily in fuels. transporting electrical energy at high
Fault ride through – Capability to remain voltage (> 35 kV) as alternating cur-
connected to the electricity grid in the rent.
case of a fault in the grid resulting in a High-voltage dc – Two-wire system for
short circuit. transporting electrical energy at high
Final energy use – Energy used by end voltage (> 300 kV) as direct current.
consumers, such as industries, com- HVAC – Acronym for heating, ventilation
mercial users and households. Does and air conditioning.
not include the energy consumption Hydrogen economy – The idea that
needed for processing fuels or power hydrogen serves as a major energy car-
plant losses. rier in the world.
Fixed charge rate – Rate of capital costs Intermediate load – Electric load that is
resulting from a given discount rate present only during 10 to 18 hours per
and a given life of an installation. day, due to the increased demand from
Forecasting error – Difference between industry, commercial buildings, and
the forecasted output from e.g. wind households.
Inertia constant – The energy stored in the Peak shaving – Decreasing peaks in elec-
rotating elements of a generating unit tricity demand by reducing demand,
divided by its nominal power capacity. using stored energy or generating elec-
Key performance indicators – Important tricity with flexible power plants.
numbers indicating the performance Photovoltaics – Dc electricity genera-
of machinery, such as the specific fuel tion directly from solar irradiation with
consumption (MJ/kWh), specific light-sensitive elements.
investment costs (€/kW), and avail- Plateau load – See intermediate load.
ability (%). Power – The capacity to perform work
Load shedding – Switching off electricity within a certain time span (joule/
users or appliances in order to balance second = watt).
electricity production and demand. Power capacity – The nameplate power
Micro cogeneration – Cogeneration instal- output or consumption of a machine.
lation intended for homes with an elec- Primary energy supply – The energy
tric output of up to a few kW, where the supply based on fuels, nuclear power,
heat released is utilised for heating the or renewables but which is not yet con-
building and sanitary water. verted into, for example, electricity or
Mineral oil – Oil from oil wells of fossil treated, such as mineral oil into petrol.
origin. Prime mover – Machine that can drive
Nominal power – The nameplate power a process by supplying mechanical
capacity of a generator or an electric energy.
appliance, often equal to the maximum Purchasing power parity – PPP, the
power capacity. buying capacity of the US$ in a certain
Non-spinning reserves – Generation country.
capacity that runs only when secondary Ramping down rate – The speed with
and tertiary reserves are needed to which a generator or prime mover
compensate for contingencies. can decrease its power output (e.g.
O&M – Operation and maintenance. MW/s).
Operational availability – The time that a Ramping up rate – The speed with which a
machine is available after maintenance generator or prime mover can increase
requirements and logistical delays have its power output.
been taken into account. Reactive power – Product of voltage and
Phase shift – The angle that two sine waves current when the two quantities are
are shifted with respect to each other. 90 degrees out of phase so that no net
Phase change of materials – Changing energy is released.
the physical condition of a material, Reliability – The probability, often
such as ice into water by melting, water expressed in the percentage of time,
into steam, or vice versa. that a machine can statistically perform
Peak factor – See crest factor. its duty.
Peak load – Either the load that occurs Renewable energy – Energy not resulting
above intermediate load, or the max- from fossil fuels or nuclear fuel.
imum load that occurs during a certain Rotating inertia – Property of a rotating
time span. mass resulting in containing energy
in proportion with the square of the Specific maintenance costs – Capital
number of revolutions per time unit. costs of carrying out maintenance per
Round-trip efficiency – Energy efficiency unit of a product, such as €/kWh.
of processes, where the energy ulti- Spinning reserve – Generating capacity,
mately ends up in the same shape as it generally expressed in MW, that is syn-
started, e.g. electricity from a battery. chronised with the grid and ready to
Shoulder load – See intermediate load. produce electricity when needed.
Simple cycle – Thermodynamic process Supergrid – See electricity highway.
where fuel energy is converted into Synchronisation – Process of ensuring
mechanical or electrical energy in a that a generator has the same frequency
single process. as the electricity grid, while the valleys
Single cycle – See simple cycle. and crests of the sine waves of the gen-
Smart Power Generation – Arrangement erator and grid coincide.
of a number of agile generating units Transmission grid – High voltage grid con-
in parallel that can undergo multiple necting generators with large consumers
starts and stops without suffering from and the distribution grids, also for the
extra wear, which results in load-inde- cross-border exchange of electricity.
pendent fuel efficiency and minimal Turn-around efficiency – See round trip
maintenance costs per kWh. efficiency.
Solar irradiation – Radiation received Unbundled power sector – Situation
from the sun at a certain location at a where electricity production, transmis-
certain time. sion, distribution and retailing are car-
Spatial impact – Space (volume and area) ried out by separate companies.
needed for installations such as a power Utilisation factor – Fraction of use of the
plant. nameplate power capacity of machinery
Specific capital costs – Capital costs per during a certain time span.
unit of a produced product, such as €/ Vertically integrated utility – Company
MWh. that produces, transports, and distrib-
Specific fuel costs – Capital costs per unit utes electricity (and/or gas, water, dis-
of fuel energy, such as €/GJ. trict heating), and sells it to end users.
POWER SUPPLY
CHALLENGES
Wind and solar energy are essential to meet energy demand and to cut
carbon emissions. But how to merge their variability into decades-old power
systems? Searching for answers, this landmark work combines empiric data
with theory of power engineering. In a truly cross-disciplinary manner, it
reaches from detailed electrotechnical analysis to system-level political and
economical issues. The award-winning author Jacob Klimstra unfolds
r ĉ FJNQBDUPGJOUFSNJĨFOUHFOFSBUJPOUPHSJESFMJBCJMJUZ
r ĉ FQPUFOUJBMPGFOFSHZTUPSBHFUFDIOPMPHJFT
r &OFSHZDPTUTBOEDPNQFUJUJWFOFTTPGEJĎFSFOUHFOFSBUJOHUFDIOJRVFT
r )PXUPCVJMEUIFQFSGFDUDBQBDJUZNJY
It is becoming widely understood that future power systems must be
FYUSFNFMZĚFYJCMF8IBUSFNBJOTVOEFSEFCBUFJTXIBUBSFUIFCFTUNFBOT
GPSĚFYJCJMJUZ8JUIDPNQFMMJOHTDJFOUJėDFWJEFODF UIJTCPPLQSPQPTFTUIBU
GBTUSFBDUJOH NPEVMBSHBTėSFEQPXFSQMBOUTBSFPěFOUIFNPTUDSFEJCMF
answer.

“The book addresses both the challenges and possible solutions for future power systems,
VTJOHDMFWFSJMMVTUSBUJPOTBOEFYBNQMFTGSPNBSPVOEUIFXPSMEĉ
using  JTCPPLJTBHSFBUSFGFS
clever illustrations and examples from around the world. This book is a great
ence for anyone
reference involved
for anyone in theinenergy
involved industry.
the energy ”
industry."
.JDIBFM8BMTI%JSFDUPS 'VUVSF(SJET &JSHSJE *SFMBOE

iĉ SPVHIIJTQSPGPVOEJOTJHIUUPFOFSHZJTTVFT UIFBVUIPSTVDDFFETUPFYQMBJOJOBOFBTZ


and understandable way the science and facts behind integrating large amounts of variable
SFOFXBCMFTJOUPFOFSHZTZTUFNT)FFYQMBJOTJOBOFMFHBOUXBZUIFVOEFSMZJOHQIZTJDTBOE
provides a lot of useful data.”
1FUFS-VOE1SPGFTTPSJO"EWBODFE&OFSHZ4ZTUFNT "BMUP6OJWFSTJUZ 'JOMBOE

ISBN 978-952-93-3634-0
ISBN 978-952-93-3635-7 (pdf)

www.smartpowergeneration.com 9 7 89529 3363 40

You might also like