The EU-25 Power Sector: a System Dynamics
Model of Competing Electricity Generation
Technologies
Erik Pruyt
Delft University of Technology, Faculty of Technology, Policy and Management,
Policy Analysis Section, P.O. Box 5015, 2600 GA Delft, The Netherlands
Tel: +31/152787468 — E-mail address: E.Pruyt@tudelft.nl
August 22, 2007
Abstract.
The main goal of this paper is to explore the transition of the EU-25 electricity generation
system towards a more sustainable system characterised by lower CO2 emissions by means of
a system dynamics model of the EU-25 electricity generation sector. In this paper, the model
and the resulting dynamics are explored by means of base case simulations, policy simula-
tions, scenario analyses and (univariate and multivariate) sensitivity analyses. Finally, some
conclusions, ex-post criticisms and directions for future research are discussed.
Keywords: Electricity Generation Technology, Europe, Transition
1 Introduction
1.1 Transition of the EU Energy System
Nowadays, the transition of energy systems towards more sustainable energy systems is a very hot
issue. The|European Commission (2007)| recently published its European Energy Policy focussed
on three important objectives: the environment (combating climate change), security of supply
(limiting the European external vulnerability to imported hydrocarbons) and economic-s
welfare (‘promoting growth and jobs, thereby providing [...] affordable energy to consumers
Much attention is paid —in that and other reports- to the transition of the elec ab
towards a sus system because (i) the electricity
for large amounts of COg-equivalent (COzeq) emi: (ii) it hi
substantially to these three objectives and (iii) ¢
more important form of energy than it is today.
Today, the transition of the European electricity system towards a more sustainable
perceived ~by the European Commission, the European electricity industry (Eurelectric) and green
NGOs alike— to be feasible because of several reasons. First, the electricity industry is believed
to possess the leverage to drastically reduce emissions over time through fuel switching, demand
reduction, efficiency improvements, and a more revolutionary transition towards a CO9-poor gen-
eration mix consisting of new renewable technologies, new nuclear ation technologies and
the use of carbon capturing and storage technologies. Many of these CO3-poor generation tech-
nologies already exist and are (almost) competitive, or are currently under development. Second,
huge investments in new generation capacity are required whether a transition to a CO-poor
generation mix is pursued or not~ and the sector has the necessary financial leverage. And third,
it is argued that there are many advantages on other dimensions related to a more sustainable
generation mix which are worth pursuing, such as improved health, lower imports of and thus a
tor is currently responsible
the leverage to contribute
n even
lower external vulnerability to- imported hydrocarbons, more growth of innovating industries and
more (local and high-tech) jobs. Higher efficiency, reduced demand and learning effects might even
keep consumer electricity prices affordable. Thus, the European Energy Policy seems to promise
good performance on all of these different dimensions —which were until recently assumed by many
to be incompatible, even opposites. And there might even be 'free lunch’ policies! Whether such
free lunches exist in the European electricity system has been explored with system dynamics
models and will be discussed in this paper.
1.2 Characteristics of the European Electricity System
very peculiar service. It is an intermediate form of energy that is currently difficult
ive to store on a large scale and needs to be generated from different other forms
of energy to satisfy electricity demand at any moment. Most European electricity is currently
generated in large centralised power plants and is transmitted and distributed through the national
grids that are supposed to become ever more interconnected and to develop —in the long term—
into a European electricity grid.
Most energy policies have in the past been made at the level of the member states, but the
relevant policy level is also (belie nifting to the European level. The European Energy
Policy shows that the EU has s icy-making ambitions in the energy/electricity domain.
There are many interactions .d) feedback effects between and especially within these
levels, which cause complex systems behaviour over time (e.g. : behaviour or boom-and-bust
behaviour), path dependence and lock-ins. The energy/elec s
(Guth ag the global climate system) are particularly characterised by the existenes of important
delays, and long average life- and response times. The average lifetime of European generation
capacity amounts currently for example to about 20-50 years and learning and lock-in effects play
on longer time scales, and other consequences” play on even longer time scales. The effects of the
energy/electricity systems should therefore not only be analyzed in the short term and medium
term, but in the long and very long term too (intergenerational time sc The (past) inertia
of the system does however not mean that it is not dynamic and that it could not drastically
change in the future: such a drastic change is precisely what is envisaged by well-considered
transition management. And the dynamic behaviour of the energy/cle tem is also multi-
dimensional in nature, directly impacting the economic, social, environmental, technological and
security of supply dimensions.
stem and affected s
les).
1.3. Consequences for Policy Analysis and Policy Design
An effective and efficient trans em towards a sustainable elec-
ion of the European electricity s
tricity system (with drastically reduced CO» emission, increased security of supply and economic
growth and more jobs) requires the design and implementation of appropriate, timely, informed
multi-dimensional dynamics on multiple dimensions is needed.
The goal of the model discussed in this paper is precisely the generation of such understanding.
Initially, the aim of this modeling endeavor was to explore the potential development of the
European wind power sector over time. But in the end, the potential dynamic development
of the entire European power sector had to be looked at because the developments of other
generation technologies are major determinants of the development of wind power itself. The
dynamic development —from the very short term to the very long term~ of the entire European
electricity sector will therefore be looked at in this paper, with a special focus on wind power.
1 Free lunch’ policies are policies *that improve some or most measures of performance without degrading others”
251).
2such as the average atmospheric life-time of CO2 which is estimated to be about 100-200 years, and is assumed
to increase global temperatures for centuries and to raise ocean levels for millennia
In view of this, a system dynami
the potential consequences of polici
simulation results are then explored and evaluated. The:
*predictions’, but rather as p.
could be used to make more robust dec
model is developed in this paper which is used to analy:
s on multiple dimensions over time. The multi-dimer
simulation results should not be
ible evolutions from which understanding might be derived which
1.4 Organization
In section] the structure of the system dynamics model will be discussed. The dynamics
model will be explored in section3|by means of a base c s, policy simulations, sc
iate and multivariate) sensitivity analyses. These multiple dimensional results
are compared to other mode
directions for future research are dis
simulatio:
and the literature. Finally, some conclusions, ex-post criticisms and
ion fl
issed in sec
2 The Structure of the System Dynamics Model
2.1 Model Boundary and Influences Considered
The focus of the model discussed here is on the potential long-term dynamics of 9 electricity
generation technologies®: gas-, coal-, nuclear-, wind-, biomass-, PV-, ‘clean coal’-, hydro- and
geothermal-based electricity generation technologies. The EU-25 policy level had been chosen at
the time of its conception because it seemed to be the most relevant policy level for the future
European electricity system. The focus on the high-level dynamic complexity of the competing
technologies means that the model is highly aggregated: individual countries, companies, power
plants, consumers, grids,... are not considered explicitly. The final time horizon of this model is
the year 2100 ~although most policies simulated are phased out before 2030- in order to analyze
the long term dynamics of these short(er)-term policies. These aggregations, simplifications and
modeling choices are acceptable only in view of the particular goal of exploring the general qual-
itative model behaviour in order to increase the understanding of the link between the structure
and long-term qualitative time evolutionary behaviour of the simulated policies/structures, not of
precise forecasts
Endogenously modelled causal influences are among else: (i) the competition between these
power generation technologies for supplying electri
generation technologies for new generation capacity to be installed; (iii) the dynamic (endogenous)
technological change of electricity generation technologies by means of experience curves; (iv) trade
in COp tradable emission certificates and tradable green certificates.
y; (ii) the competition between these power
pacity;
) angregated and s
(ii) the societal value
and grid integ)
mnplified way) include: (i) the maximum potential wind power
em and related to that, the public acceptance of wind power; (iii) siting
(iv) the increasing need for backup generation and storage capacity
with higher s iable wind power output in the total electricity system, decreasing the
price and increasing the costs of wind energy; (v) the electricity demand and related to that
demand side management and rational energy use; (vi) and the European import dependence and
the potentially resulting energy system uncertainties and stresses
Many other causal influences are thought to be important but are nevertheless modelled exoge-
nously because their endogenous inclusion would require additional structures which would only
draw the attention away from the issue of interest, namely the potential development of European
wind power (and other power generation technologies) in order to decrease the contribution of the
electricity sector to climate change. These exogenous influences are neverthele:
sensitivity analyses in order to assess their potential impact on the system dynam’
Included exogenous variables are among else: (i) the general European economic development
(which should actually be rendered partly endogenous); (ii) the evolution of the European energy
s varied in scenario
model.
and
3See (Pruyt 2007b) for a detailed description of the model structure and multi-dimensional dynamics
(iv) the public acceptability of wind turbines ss 5
(v) the degree of effective liberalization and competition; (vi) and the public support for nuclear
power. Many related issues, influences and structures are not integrated in the model (see 28).
2.2 Feedback Loops — Feedback Loop Diagrams
The model contains many important feedback loops such as: positive experience-cost loops © ,
a positive wind power potential expansion loop @ , negative maximum potential loops, positive
profitability loops ¢ , positive competition for generation loops ¢ , negative decommissioning loops
©, negative diminishing profitability loops, positive generation cost redu ©, a positive
capacity expansion of the wind power construction industry loop @ , positive new capacity required
loops @, positive experience improvement loops & , negative backup and storage loops¢), a negative
fuel import dependence loop, negative expected TGC and TEC cost loops, a negative demand
elasticity loop~. These feedback loops are discussed in detail in section 1 of the appendix. The
simplified stock-flow diagram in figure [I] (15) also shows some important feedback loops for all
generation technologies: the positive experien loops and the positive profitability loops.
So, the model includes learning and experience effects, cost expectations, expectations of prof-
itability of generation technologies potentially to be installed, effects of (dynamic) maximum
potentials, effects of industrial expansion, competition effects between generation technologies
and demand reductions, price-demand elasticity effects, intermittence effects (requiring sufficient
backup or storage) and fuel import dependence effec
2.3. Stocks and Flows — Stock-Flow Diagrams
The stock-flow diagrams of this model are displayed and discv
appendix. Figure{l|shows a ‘summary’ stock-flow diagram of some of the most important feedback
loops for a generic technology i. The main stock variable for any particular technology is of course
the capacity installed of that particular technology i.
3 Exploring the Dynamics
It takes a rather time and energy consuming iterative exploration proc
and interpret such dynamic models: many alternative structures and polici
and simulated and many scenario analyses, uni-variate and multi-variate sens tivity analy
to be performed. This s also wary complex anid there are too many wuncertainti
and simplifications in order to use this model for point or trajectory predictions or to pay much
attention to numerical sensitivity", The general behaviour, behaviour mode and policy s
are therefore of interest. The general behaviour of a structural base cas
to build, test, simulate
need to be modelled
need
sction, followed by a discussion of the general behaviour of se
and sensitivity analyse
eral policies, scenarios
3.1 Dynamics of the Base Case Scenario
Settings of the Base Case Scenario In the base case (BC) scenarid®), all progress ratios
are equal to 85% except the progress ratio of PV power capacity which equals 75%. The initial
wind power capacity factor is 1/3, the initial capacity of the wind power construction industry is
10GW, and the annual maximum percentage growth wind power construction industry is 30%. The
maximum wind power potential lookup is displayed in figure(a and the siting and other investment
“The work presented here was the immediate cause for the research on System Dynamics and Uncertainties’ as
presented in (Pruyt 2007a).
®'The base case discussed here should be
the link bet
nas a set of basic structural assumptions in order to better understand
1 behaviour. It is not a ‘Business As Usual’ or reference scenario.
en model structure and mo«
+ Profitability 4 —— electricity price
4
technology i ea geaealy of ecole ie
Figure 1: Summary stock-flow diagram of some of the most important feedback loops for a generic
technology i
's lookup in figure 2b. These lookup functions are nothing but guestimates and will therefore
be subjected to sensitivity analyses.
The initial marginal cost of wind power capacity 2006 is taken to be 900ME/GW, that of gas
capacity 500ME/GW, of coal capacity 1100ME/GW, of nuclear power capacity 2000M€/GW, of
clean coal capacity 5000ME/GW, and of new PV power capacity 6000ME/GW.
The lifetimes of these technologies are taken to be 20 years for wind power, PV and geothermal
power gencration technologies, 30 years for gas, coal, clean coal and biomass power generation
technologies, and 40 years for nuclear and hydro power gencration technologies.
The initial average fuel cost of gas based gencration is taken to be 15419€/GWh, and of coal
based generation 6665€/GWh. The additional fuel cost clean coal on top of the fuel cost of coal
based generation is 10%, but there are no additional carbon capturing and storing costs. Later on,
additional carbon capturing and storing costs will be added and subjected to sensitivity anal
cost of biomass is 80% of the normal cost if there is no biomass compared to
its maximum potential, 150% the normal biomass fuel if the biomass capa i
maximum potential, and even 500% the normal fuel cost if the capacity serious
maximum potential. Normal fuel costs are incurred here when 50% of the maximum potential
is reached. The average nuclear fuel cost is initially taken to be €2000/GWh, but there is an
sts of €1000/GWh. All market technologies also incur
The average fuel
y overshoots the
additional future nuclear waste storage
s of €8000/GWh.
The initial specific CO emissions of gas based gencration amount to 400tCO2/GWh, the
initial specific CO emissions of coal based generation to 800tCO2/GWh, and there is a percentage
CO» emissions of clean coal of 10% of conventional coal emissions.
In the base case, the GDP growth EU trend is taken to be 3% per year (without a randomiser),
the electricity intensity of the EU growth is taken to be 63% and there is a price elasticity of
the elec demand of 15%. There is no annual percentage decrease of the electricity demand
by forced DSM and REU. The electricity price structure used here is equal to 1/3 the apriori
other variable cos
7G wanton EURO Ww EUROAT
Nem Noe =
oa Vas zines ingot le | Simms pwh7TOe? man]! =| ent Seg
OK | Cw Purte| Cust | Crh) Chur ence | fbicul Cored | OR | Cau Parts| ChwAbParts| Cusfiel| Cau Faleance | Ral>Cs] Cored
Figure 2: The maximum wind power potential lookup (left) and the siting and other investment
costs lookup (right)
electricity market price plus 1/3 the aposteriori electricity market price plus 1/3 the aposteriori
electricity market price of the previous period. And there is a general electricity price markup of
maximum 5% in case of shortages.
There is no real hydrogen or storage breakthrough hence no additional growth rate electricity
demand by hydrogen breakthrough and no storage capacity for intermittent generation— and no ini-
tial investment in clean coal capacity. The maximum biomass power capacity potential is 500GW,
the initial maximum potential PV capacity 1000GW, and the maximum potential hydro capacity
262.88GW. In the base case, there are no wind power capacity investment subsidies, no initial
investment in clean coal capacity, no percentage TGCs required, the TAX/TEC price is €0/tCOs,
there is no public support for new nuclear power and the greenness of the societal value system is
neutral, hence not specifically favouring wind power.
In the base run of the base case there is no additional decommissioning nuclear capacity or
rated nuclear phase-out, and there is no renewed public support for new nuclear power.
The description above makes clear that the base case is actually an extreme scenario in that it
does not contain many climate change policies to be expected or currently implemented. At most,
it could be seen as a reference run of what would happen without any climate change polic
which is given the current context quite unrealistic. The base case is therefore not to be seen as a
’Business As Usual’ (BAU) scenario. The moderate climate change scenario discussed on page [7]
is currently more like a BAU scenario for the EU-25. This base case is therefore not really useful
for policy relevant questions. But it is useful for assessing the overall dynamics of the model, and
for assessing the influence of the policy measures. And even if the base case would be an adequate
BAU run, then it should still be kept in mind that the model is nothing but a micro theory or
semble of structural assumptions. Many of these ‘assumptions’ will be varied in this and
an ¢
following subsections.
Simulation of the Base Case Scenario It should be kept in mind that the dynamics discussed
and the following paragraphs are not projections, predictions, or foresights. They
are first of all simulations to increase the understanding of the structure and the dynamics of
This is one of the reasons why general time
evolutionary dynamics are generated in the rest of this section instead of any (precise) numerical
outcomes: the dynamics of the simulations will mainly be discussed by means of figures displaying
the evolution of key variables over time
The BC electricity demand increases steadily. The normal electricity supply -the amount
of electricity potentially generated— periodically falls short of demand in the second half of the
century, but the absolute maximal potential electricity generation allows to cover the electricity
demand at all times, even in tight times. This rily so in case of strongly fluctuating
demand.
foreca:
in this
the model during the iterative modelling proce:
The total CO, emissions from electricity gencration continue to increase to about the sixfold
of current emissions. The reason for this increase is to be found in the increasing electricity de-
mand/supply combined with (more or less) constant specific CO» emissions of electricity generated
resulting from the combined influence of a decreasing fraction of CO2-poor generation technolo-
an increasing fraction of gas based gencration and decreasing specific CO» emissions of coal
and gas based generation due to efficiency gains and learning and experience effects.
The fraction of COs-poor electricity generation decreases from 50% to about 20% (because
of the continuously decreasing nuclear generation), whereas the fraction of coal remains rather
stable, the fraction of gas-based electricity generation continuously increases, and the fraction of
renewables increases until the middle of the century after which it slightly decreases.
Initially, the increase of the renewable generation fraction is mostly attributable to the rapid
increase of the biomass fraction among else due to a high premature conversion of coal to biomass
capacity. From 2022 on, the increase of the renewable fraction could be attributed mainly to the in-
creasingly important fraction of PV electricity generation. The decreasing renewable fraction after
2050 could be explained by the maximum potentials reached of several of these renewable power
options, whereas other potentials remain untapped in the absence of policies, for instance that of
wind power (because of the intermittence requirement and the high amounts of PV capacity).
The enormous capacity installed of the PV type leads to less then 15% of generation because of
its particular low but dynamically increasing- capacity factor. Decisions about capa to be
installed are very important because they determine for a very long time, what technologies will
be used for generation purposes and what technologies will be further invested in due to learning
effect dynamics. The enormous capacity installed of the PV type additionally leads directly to
lower wind power capacity.
The annual investments suggest decennial inv
more pronounced because of the real-world comm
(discontinuous) amounts of generation capaci
ment s which might in reality be even
ioning and decommissioning of large fixed
3.2 Policy Analyses
Three sets of climate change policies will now be introduced. These three base policy sets are
compared with the base run of the BC, mainly by means of figures showing the evolution of
important variables over time for the BC and the 3 climate change policy sets.
Description of the 3 Policies: The three climate change polic ssed here are actually
sets of diverse policies assumed to be applied to the entire EU-25 POLI policy set is a
set of moderate climate change policies. The base POL? policy set is a set of medium to strong
climate change policies. And the base POL3 policy set is a set of very strong climate change
policies. Many more policy sets could be simulated with this model, but only these three ~and
several variants~ will be discussed here.
« In the base POLI policy set ~a set of moderate climate change policies~ there is
— no annual percentage decrease electricity demand by forced DSM and REU,
— no accelerated/ additional decommissioning of nuclear capacity as currently for
many European countries
— no renewed political/public support for new nuclear power,
— no initial investment in clean coal capacity,
— arising greenness of the societal value system, rising from 0% in 2006 to 10% in 2020
and remaining 10% thereafter,
— a public support for wind generation of 10% above neutral (1.1 in the model),
—a wind power capacity investment subsidy of 10% until the year 2020 after which it
drops to 0%,
— a TGCs system with a percentage of TGCs required increasing linearly from 0% in 2005
to 10% in 2020 dropping to 0% afterwards, a minimum TGC price rising from €0/TGC
in 2005 to €10/TGC in 2020, and a maximum TGC price rising from €10/TGC to
€50/TGC in 2020,
— and a deterministic TAX/TEC system increasing linearly from €0/tCO» in 2005 to
€50/tCOs in 2100.
e In the base POL2 policy set —a set of medium to strong climate change policies~ there is
— an annual percentage decrease of the electricity demand by forced DSM and REU of
1% per year from the year 2006 until the year 2012, as foreseen in a recently proposed
European directive,
— no accelerated/ additional decommissioning of nuclear capacity,
— no renewed political/public support for new nuclear power,
— arising greenness of the societal value system, rising from 0% in 2006 to 10% in 2020
to 30% in 2050 and 50% in 2100,
— a public support for wind generation of 10% above neutral,
ment subsidy of 20% until 2030 and 0% afterwards,
— an initial investment in clean coal capacity of 2GW/y in the years 2010 and 2011,
— a TGCs system with a percentage of T@Cs required increasing linearly from 0% in 2005
to 20% in 2020 and 0% afterwards, a minimum TGC price increasing linearly from
€0/GWh in 2006 to €20/GWh in 2020, and a maximum TGC price increasing linearly
from €20/GWh in 2006 to €50/GWh in 2020,
— and a TAX/TEC increasing linearly from €0/tCOz in 2006 to €100/tCOz in 2100.
— a wind power capacity in
e In the base POL3 policy set ~a set of very strong climate change policies~ there is
— an annual percentage decrease of the electricity demand by forced DSM and REU of 1%
per year —not only until 2012 as forescen in the recently proposed European directive—
but until the year 2030,
— no accelerated/ additional decommissioning nuclear capacity,
— no renewed public support for new nuclear power,
sing from 0% in 2006 to 10% in 2020
— arising greenness of the societal value system,
to 30% in 2050 and to 50% in 2100,
— a public support for wind generation of 10% above neutral,
— a wind power capacity investment subsidy of 20% until 2030 and 0% afterwards,
— an initial investment in clean coal capacity of 4GW/y in the years 2010 and 2011,
—a TGCs system with a percentage of TGCs required increasing linearly from 0% in
2005 to 20% in 2020 and 50% by 2030 and afterwards, with a minimum TGC price
increasing linearly from €0/GWh in 2005 to €20/GWh in 2020 and afterwards, and a
maximum TGC price increasing linearly from €20/GWh in 2005 to €50/GWh in 2020
and afterwards,
— and a TAX/TEC increasing linearly from €0/tCOz in 2006 to €200/tCOz in 2100.
Comparison with the Base Case Figure |3| (#9) shows that the electricity demand differs
strongly between the BC, POLI, POL2 and POL3 policy sets. The reason for this marked differ-
is that DSM and REU play an inereasingly important role in these respective elimate change
. The differences in electricity demand and the differences in specific CO emissions of
the electricity generated (see figure [5] ({10)) account for the very different evolution of the CO2
Graph for electricity demand
°
2005 2010 2018 2020 2025 2030 2040 2045 2050 2085 2060 2065 2070 2075 2080 2085 2090 2095 2i00
Time (Year)
electricity demand : BC Gwhy
electricity demand : POL1 --~ - sosceegeneeeeegieeeeeegereeeegen GWhy
cleetricity demand : POL? + — —¢— 3 —3— 3 3 pe GwWhy
electricity demand : POL3 - <--— GWhy
Figure
The electricity demand in case of the BC, POL1, POL2 and POL3 polici
Graph for percentage CO2 emissions base 2006
oo figeelige:
2005 2010 2015 2020 2025 2030 2035 2040 2045 2080 2055 2060 2065 2070 2075 2080 2085 2090
Time (Year)
percentage CO2 emissions base 2006 :
percentage CO2 emissions base 2006 :
percentage CO2 emissions base 2006 :
percentage CO2 emissions base 2006 :
Figure 4: The percentage CO emissions in function of the base year 2006 in case of the BC,
POLI, POL2 and POL3 policies
10
Graph for specific CO2 emissions of electricity generated
ee ee a
80
40
0
2005 2010 2015 2020 2025 2030 2035 2040 2035 2050 2055 2060 2085 2070 2075 2080 2085 2090 2095 2100
Time (Year)
specific CO2 emissions of electricity generated : BC 1C02/GWh
specific CO2 emissions of electricity generated : POLI----
specific COZ
specific CO:
dope *CONGWh
eCOUGWh
iCOGWh
smissions of electricity generated : POL3— -4.-—- 4 -—--4-— === = a
Figure 5: The speci
POLS policies
CO emissions of electricity generated in case of the BC, POLI, POL2 and
Graph for total CO2 emissions
200B
180 B
160 B
1408
1208
1008
2008 2010 2015 2020 2025 2030 2035 2040 2045 2080 2055 2060 2065 2070 2075 2080 2085 2090 2095 2100
Time (Year)
total CO2 emissions : POL 1c02
total CO2 emissions : POL2 2. a oe eee ee a
total CO2 emissions : POL3 —g— 3-3 9 ee 1?
Figure 6: The total cumulative CO» emissions of el
POL2 and POL3 poli
tricity generated in case of the BC, POLI,
ies
ll
Graph for fraction of renewable generation of total electricity generation
wt eet ig
4
0.64 na =
m3
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2085 2060 2065 2070 2075 2080 2085 2090 2095 2100
Time (Year)
fraction of renewable generation of total electricity generation : BC Dmal
fraction of renewable generation of total electric POLE----2--- Dmal
fraction of renewable generation of total electricity generation : POL23 — —3— Dmal
fraction of renewable generation of total electricity generation : POL3.-4-—-4-—-~ - - Dmal
Figure 7: The fraction of renewable generation of total electricity generation in case of the BC,
POLI, POL2 and POL3 policies
emissions (sce figure {Jj (59). The BC results in an exponential increase of the annual CO emis-
sions to about 6 times the 2006 emissions in 2100. The moderate climate change policy (POL1)
leads ~after an initial increase followed by a decrease and again an increase~ to emissions of about
16% above year-2006 emissions. The medium climate change policy (POL2) leads to gradually
decreasing CO, emissions to less than half the 2006 emissions. And the very strong climate change
policy (POL3) results in a more rapid decrease of CO» emissions to less than a fifth of 2006 annual
emissions.
However, the annual emissions are not as important as the cumulative CO» emissions displayed
in figure(6|(i10) for POLL, POL2 and POL3. There it could be seen that these different paths make
a huge difference in terms of behaviour of the accumulated emissions which eventually partially
make up the atmospheric concentration.
The decreasing average specific CO emissions per GWh generated of the three climate change
policies are the result of the increasing fractions of renewable generation (sce figure{7/ (#11)) —most
notably biomass gencration (see figure 8] (HI2))- and CO»-poor generation in general (see figure|9]
(112) —including clean coal generation (sce figure [10] (#13))-, and the decreasing fraction of gas
and especially coal based generation (see figure {Tl (uT3)).
Wind power generation makes up ~at least in the base runs of the BC, POL1, POL2 and POL3-
only a relatively tion of the renewables and total generation b
factor and low stalled. The peak is mainly caused by TECs and TG
stment subsidies and the’subsequent decrease is mainly caused by an overshoot of the
capacity ins
intermittent
of wind power
It is inter
e of a low
direct inv
alled above the dynamic maximum potential, the negative impact of a high fraction of
used by the increasing PV capacity installed~ on the further development
apacity, and the development of other CO» poor generation technologies.
ing to sce that the non-subsidised BC wind power capacity installed in absolute
terms~ increases above that of heavily subsidised POL3 by the year 2045 and reaches about the
same levels of capacities installed of the POL2 and POLI policy sets. The underlying reason is
12
Graph for fraction of biomass of total electricity generation
2005 2010 2015 2020 2025 2030 2035 2040 2045 2080 2055 2060 2065 2070 2075 2080 2085 2090 2095 2100
‘Time (Year)
fraction of biomass of total electricity generation
fraction of biomass of total electricity generation
fraction of biomass of total electricity generation
fraction of biomass of total electricity generation
Figure 8: The fraction of biomas
and POL3 policies
s generation of total generation in case of the BC, POL1, POL2
Graph for fraction of CO2 poor generation of total electricity generation
ars
2005 2010 2015 2
2025 2030 2035 2040 2045 2050
Time (Year)
2060 2065 2070 2075 208 2085 2090 2095 2100
Dmal
- 2 spe Dal
3 a Dl
Saeed Dinal
fraction of CO2 poor generation of total electricity generation
fraction of CO2 poor generation of total electricity generation
fraction of CO2 poor generation of total electricity generation
fraction of CO2 poor generation of total electricity generation : POL3-- ¢-—- 4-—-4----—--4—-=
Figure 9: The fraction of CO poor generation of total el
POLI, POL? and POLS policie
ricity generation in case of the BC,
13
Graph for fraction of clean coal generation of total electricity generation
2005 2010 2015 2020
fraction of clean coal generation of total electricity generation
fraction of clean coal generation of total electricity generation
fraction of clean coal generation of total electricity generation
fraction of clean coal generation of total electricity generation :
2025
2045 2050 2055 2060 2065 2070 2075 2080 2085 2090 2095 2100
‘Time (Year)
Be Dmal
POLE sepa gued ge ganas — Dmal
POL23— 3 3 Dl
POLB -4-— =a gg Dima
Figure 10: The fraction of clean coal generation of total electricity generation in case of the BC,
POLI, POL2 and POLS policies
Graph for fraction of coal based generation of total electricity generation
SS
aed Fey
2005 2010 2015 2020
fraction of coal based generation of total electricity generation
fraction of coal based generation of total electricity generation
fraction of coal based generation of total electricity generation
fraction of coal based generation of total electricity generation
Figure 1: The fraction of coal generation of total electricity generation in ¢
POL2 and POL3 poli
202
ies
aires 5 et
2045 2050 2085 2060 2065 2070 2075 2080 2085 2090 2095 2100
Time (Year)
BG Dmal
POL Begins genre teen teeneeebineen ~ Dmal
POL23 — 3 — 3 9 ml
POLS «2 — = yong peg ge gp Dial
» of the BC, POLA,
4
the more substantial growth and total installed capacity
The cumulative private investments from the year 2006 on are displayed enormous and still
lead to potential generation shortages if the capacities are only used for generation within their
normal capacity factor limits. These generation shortages slightly impact the electricity price:
the BC leads to slightly decre ices, that the POLI climate change policy set
leads to slightly increasing clec and that the POL2 and POLS sets lead in the end to
almost doubling and tripling electric Three main reasons for this are (i) the choice for
(initially) more expensive generation technologies than would have been the case without climate
change policies, (ii) the additional price paid for TGCs, CO2 TECs and taxes on remaining CO.
emitting generation, and (iii) the omission of external costs or potential climate change related
costs.
y.
3.3 Scenario/Policy Analyses
Gas Price Scenarios: Two gas price scenarios ~a high and a low gas price scenarid®- are —by
means of the average fuel cost gas as percentage of initial fuel cost lookup— applied to the BC,
POLI, POL2 and POL3 poli . In the high gas price scenario, the gas fuel cost expressed in
percent of the initial fuel cost grows linearly from 100% in 2005 to 150% in 2020 to 180% in 2050
and to 200% in 2100. In the low gas price scenario, the gas fuel cost expressed in percent of the
initial fuel cost decreases from 100% in 2005 to 70% in 2050 and to 50% in 2100.
These gas price scenarios most heavily impact the BC simulations because they are charac-
terised by a higher electricity demand and a higher fraction of gas based electricity generation
than the other polic
The total cumulative investment costs are much lower in case of low gas prices
the low marginal investment cost of new gas capacity. The total cumulative priv
are about €10!? lower (on a total amount of about €8 billion) in the low gas price scenario of
the BC simulations than in the high gas price scenario. This leads to lower el y prices and
consequently more electricity demand in case of low gas prices which leads to almost 2000GW
more cumulative new capacity installed in case of low gas prices.
The impact of these gas price scenarios on the fraction of CO, poor electricity generation is
also more pronounced in the BC and the POLI set, which leads to a marked difference in total
cumulative CO emissions. It is interesting to sce in the BC runs that both the lower and —
especially~ the higher gas price scenarios lead to lower total cumulative CO» emissions than the
normal price scenario. In the low gas price s gas-based generation substitutes coal-based
generation and therefore leads to lower specific emissions than in the normal price scenario. And
in the high gas price scenario, this is due to the much lower fracti as-based generation,
the higher fraction of clean coal based generation, the lower demand, and the higher renewable
fraction foremost due to the increased biomass based generation and to a lesser extent wind
power generation. The fractions and absolute values of wind power capacity are only impacted in
the BC simulations in the second half of the century. A possible reason for this is the need for
flexible generation and/or storage capacity to compensate for the intermittent character of wind
power. This will be explored in the following paragraph.
Nevertheless could it be concluded that the higher and lower gas price:
ent technology development and investment paths and hence to a different generation mix.
also means that the gas price could (potentially) be used for policy-making purpo:
e investments
enario
lead in this model to
diff
Tl
Real Hydrogen or Storage Breakthrough Scenarios: The base BC, POLI, POL2 and
POL3 policy sets are not characterised by a real breakthrough of hydrogen or cheap electricity
storage technology. The impact of such a hydrogen (or storage) breakthrough in the year 2020
will be looked at in this paragraph. First, a breakthrough without an additional increase of the
electricity demand and then a breakthrough with an additional growth of the electricity demand
6Fluctuating gas prices have also been simulated and lead to intermediate results.
15
by the hydrogen breakthrough for example caused by the gradual switch from fossil fuels to
hydrogen by road transport~ is looked at.
Without additional growth of the electricity demand, the model shows a strong impact of a
real hydrogen or storage breakthrough in 2020 on the penctration of wind power from that year
on. The main rei ourse that a higher fraction of intermittent electricity capacity
requiresveither backup generstion capasityor significant cheap storage capacity: In. tlie model,
new intermittent capacity is disfavoured and flexible non-intermittent generation is favoured with
increasing fractions of intermittent capacity before a real hydrogen or storage breakthrough. But
is constraint ceases to exist once cheap hydrogen or a mass-storage technology becomes available.
ive storage could from then on form a buffer between the intermittent supply and variable
demand.
This scenario leads to a substantial increase of the fraction of wind generation: on average
about a quarter higher in 2030, about twice as high in 2050 and even higher in 2100. The same
goes for the fraction of wind power capacity installed of total capacity installed depicted in figure.
In the case of a cheap real hydrogen or
in’ the long to very long-term without stringent climate change policies:
electricity deitiand as muchshigherin stich cages’ which ‘mieana amore absblute capacity additions
and hence more sales.
There is only a small impact on the fraction of CO2 poor generation, the specific COz emissions
and the annual total CO emissions. The cumulative impact on total cumulative CO emissions
—which is what matters in : s small, but not negligible.
The impact on the fraction of renewable generation of total electricity generation is slightly
bigger, between about 5 and 10%. Thi .d by a substantial increase of the fraction of wind
power and a decrease of the fractions of gas, biomass, coal (only in the BC) and especially of clean
coal.
It also leads to marginally higher total cumulative private investments and a slightly higher
total cumulative new capacity installed since wind power requires more capacity to be installed
because of the lower capacity factor. So, the breakthrough of cheap storage or hydrogen technol-
ogy without any additional demand leads ~at least in this model- to a shift in technology.
electricity demand falls to 0%, do
but also to a strongly increased elec
electricity sector i
enario lead to slightly different specifi
‘ y demand such that the CO» emissions of the European
-while at the same time most likely decreasing in other sectors such as
It also leads to substantially higher amounts of cumulative new capacity installed and of cumula
private investment:
Furthermor
cenario initially leads to higher fractions of wind power capacity and wind
ration, subsequently turning into fluctuating fractions, which combined with the sig-
d demand lead to much more wind power capacity installed —balancing on the
ctions therefore also change. Most notable among these
s are the significant increases of the fractions of gas generation and of clean coal generation,
ignificant decrease of the fraction of biomass generation. This leads contrary to the same
scenario without additional demand- to significantly lower fractions of CO.-poor and renewable
generation. Balancing at the limit of the dynamically increasing maximum potentials also leads
to slightly fluctuating electricity prices and to higher generation shortages.
The general conclusion of this hydrogen scenario analysis is that wind power capacity develop-
ment is extremely sensitive to a real breakthrough of hydrogen (or storage) technology, especially
when this leads to an increase of the electricity demand, at least if intermittence is punished.
GDP Scenarios: The influence of the GDP growth will be looked at in this paragraph. In
the base runs, the GDP growth amounts to 3%. Here, two additional scenarios per policy set are
16
compared to these base runs, more precisely one with a high GDP growth of 4.5% and one with a
low GDP growth of 1.5%. The model also allows the assessment of the impact of fluctuating and
random electricity demand increases, which are not explored here.
The electricity demand differs significantly between these scenarios. The CO» emissions ex-
pressed as a percentage of 2006 emissions therefore also differ substantially and, in spite of the
smaller difference in specific CO. emissions per GWh. The same goes for the total cumulative new
capacity installed and the total cumulative private investments.
These scenarios actually lead to bigger differences in the fractions of renewable generation than
the fractions of CO,-poor generation (the growth of the GDP is actually —per policy set~ inversely
proportional to the fraction of renewables), because of the major impact on the fraction of clean
coal.
The growth also has an enormous influence on the absolute amount of wind powe
installed and a smaller influence in relative terms. These simulations are again characterised by
intermittence penalties and the lack of a storage or hydrogen breakthrough which partly
the overshoot and subsequent depression (boom and bust) of relative wind power capa
wind power generation in most runs.
The GDP growth also significantly influences the fractions of gas-based gencration. The growth
does not have a significant influence on the fraction of coal based generation because all scenarios
applied to the policies (except for the BC policy
of coal based generation.
It could be said that higher GDP growth initially leads ~at least in the case of POL1, POL2
and POL3- to higher fractions of biomass until the maximum potential is reached and the fra
of biomass decrease.
In the GDP scenari
slightly in the POLI cas
base policy runs. Prices
growth rates of GDP are, the lower the electricity prices in the model are.
The scenarios discussed here do not include a decoupling of the growth of the GDP and the
growth of the electricity demand by means of a decreasing electricity intensity of the EU growth.
Without such decoupling of the growth of the electricity demand from the economic growth, it
could be concluded that the growth of the GDP has a major impact on many dimensions
the wind power capacity installed.
apacity
explains
y and
ct) lead to a substantial decrease of the fraction
, the electricity price deer ightly in the BC poli ‘
doubles in the POL2 case and triples in the POL3 case as it did in the
in the various scenarios diverge only slightly. The lower the modelled
Combined Policies/Scenarios: Many other scenarios/policies and combined sc
might be simulated as well. Cross-impact assessment might also be used to
den/discontinuous occurrence of events ~such as the outbreak of a major international
sudden political interference- or sudden (r)evolutions. This also helps the in/
to such events and hence the robustness. The sudden addition of politically forced new nuclear
capacity will be looked at later. Space precludes a full discussion of such additional analyses here.
onarios/policies
s the sud-
onflict or a
3.4 Univariate Sensitivity Analyses
Sensitivity Related to the Maximum Wind Power Potential: The sensitivity related
to the maximum wind power potential could be tested by changing the maximum wind power
potential lookup function, or by changing the —specifically for such sensitivity analyses added—
maximum wind power potential lookup factor (which is 1 in the base runs), thus changing the gap
wind power potential” function.
A doubling of the maximum wind power potential without any hydrogen breakthrough— leads
to a higher absolute amount of wind power capacity installed and higher fractions of wind capacity
installed and generated. It also leads to somewhat higher fractions of renewables and CO poor
generation. But it only leads to slightly lower fractions of gas, coal, biomass, and clean coal
7 (maximum wind power potential lookup factor * maximum wind power potential lookup - wind power capacity
installed) / (maximum wind power potential lookup factor * maximum wind power potential lookup)
17
based generation, and to slightly lower relative, total and cumulative total CO, emissions. So
without a hydrogen breakthrough there is not much change except for the higher amounts of wind
power capacity and generation. A reason for this might be the fact that, here, wind power is still
constrained by the backup generation or storage capacity required.
But if a hydrogen breakthrough (or another evolution removing the backup generation or
storage capacity constraint) takes place, then a doubling of the maximum wind power potential
substantially increases the wind power capacity installed, increases the fractions of CO» poor and
renewable generation, decreases the fractions of gas based generation, of clean coal generation and
of biomass gencration. It also substantially decreases the percentage and cumulative emissions
to such an extent that other emission paths (and thus atmospheric concentrations) are actually
reached. More wind power capacity also means more cumulative investments and more cumulative
new capacity installed.
Another potential source of sensitivity related to the maximum wind power potential are the
siting and other investment costs. The sensitivity of the model to changes in this variable is
explored by means of the specially added variable siting and other investment costs lookup factor
which is 1 in the base runs. Two variants per poli are simulated here, doubling and halving
these siting and other investment costs. This only has a small influence on the wind power capacity
installed and the fraction of wind power generated, and almost no influence on the other variables
in the model. Completely other siting costs functions might however have a bigger impact but
were not tested here.
Sensitivity Related to the Maximum Biomass Potential: In previous analyses, the abso-
lute maximum biomass power capacity potential was taken to be 500GW which might actually be
too high for the EU-25. The question explored here is whether lower maximum potentials would
lead to very different dynami
If biomass is only used for electricity generation, then [Bricsson and Nilsson (2006)| s the
potential EU-25 biomass supply —using a resource-focussed approach- to lie between 4.1 and 17.2
EJ/y -compared with the biomass target of 5.6 EJ/y of the 1997 EC White Paper on Renewables—
which comes down to a maximum power potential of about 76GW to 318GW knowing that 1EJ
= 1/3.6 10°GWh and assuming an efficiency of 35% and capacity factor of 60%.
estimates —based on other studies~ the European biomass potential to amount to 8.9 EJ/y —or
LeaGw. assuming an efficiency of 35% and capacity factor of 60%. And the [European Energy|
[Agency (2005, p2)|estimate: (preliminarily) that ‘the potential of environmentally-compatible pri-
mary biomass for producing energy could increase from about 180 Mtoe in 2010 to about 300
Mtoe in 2030’, equivalent! to a maximum EU25 biomass power capacity potential of 159GW in
2010 to 265GW in 2030. Given this information, it seems reasonable to explore absolute mazimum
biomass power capacity potentials of 160GW and 280GW
The biomass capacity installed in the POLL, POL2 and POL3 policy sets grows —in a slightly
oscillatory fashion- to reach the maximum potential after 2025 and 2050. The BC policy set is
influenced but does not reach the maximum potential at all. The cumulative total CO» emissions
between 2006 and 2100 significantly increase with the lower biomass potentials. The difference
between the base POLI run and the 160GW POLI scenario run even amounts to about 20GtCO.
or more than 15% of the total cumulative emis This increase is mostly due to a decreasing
fraction of biomass based generation, an increasing fraction of gas based generation, slightly higher
coal based generation, much higher clean coal based generation and a somewhat higher fraction
of wind power generation. And such an increased fraction of wind power generation requires a
considerable additional amount of wind power capacity to be installed. The rest of the variables
in the model (electricity price, cumulative investments, et cetera) only change slightly. So it could
be concluded that the model outputs of the CO» emissions and wind power capacity installed are
actually rather sensitive to the maximum biomass potential.
‘ions
pas
Sconverting Mtoe to GWh by multiplying by 1/(8.6 * 10-5) and assuming an efficiency of 35% and a capacity
factor of 60%
18
Sensitivity Related to the Maximum PV Potential: The mazimum potential PV capacity
in this model is a dynamic maximum PV potential, although there is no truly endogenous dynamic
link between the PV module and the rest of model. This maximum potential is in the limit almost
double the initially provided initial maximum potential PV capacity. The ‘correct’ value of this
initial maximum potential PV capacity is really uncertain because the ultimate maximum potential
depends on many uncertain and unknown factors: the space ultimately available for PV pan
the relative price of the PV modules, potential technological (revolutions (Thin-Film Solar Cells)
EU-25 support, the dynamics of other generation technologies, et cetera. It is therefore nec
to explore the sensitivity related to this initial maximum potential PV capacity. In the base runs,
its value was taken to be 1000GW.
Here, a ten times lower value (100GW) and a two times higher value (2000GW) are first of all
simulated. This has of course a very serious influence on the PV power capacity installed and the
fraction of PV of total electricity generation. But in this particular form of the model, it also has
a very important influence on the wind power capacity installed and the fraction of wind power
generation, on the fraction of gas based generation, on the fraction of clean coal generation, and
a somewhat weaker influence on the fractions of CO» poor generation and renewable generation
more wind but less PV generation and vice versa~ and a weaker influence on the fraction of
coal based generation. This also means that there is a small influence on the (cumulative) CO>
emissions.
The total cumulative private investments are consequently much lower because the marginal
investment cost of PV power is relatively high, and the total cumulative new capacity installed is
also much lower because the capacity factor of PV power is relatively low which means that much
more capacity is required to satisfy the electricity demand.
Two influences might at first sight seem surprising: (i) the very strong influence of the initial
maximum potential PV capacity on the absolute amount wind power capacity installed, and (ii) the
negative influence of the (initial maximum potential) PV capacity on the CO» emissions (less PV,
less CO2 emissions). Both influences are the result of two modelling choices (and their combina-
tion) in the model. First of all, PV power generation is considered to be distributed /decentralised
‘must-run’ generation. And second, it is directly linked to wind power capacity by the requirement
that —unless there is a hydrogen breakthrough— there is a maximum percentage of intermittent ca-
pacity. Now, this influence (corresponding to my initial mental model) is rather strong in this
model whereas it might actually even be very weak in the real world. It might therefore be con-
sidered to weaken this requirement or keep the requirement but uncouple wind power and PV
power.
Sensitivity Related to the Clean Coal Power Costs: Many of the previously discussed
simulation runs —especially those of the POL3 policy set- show enormous increases of clean coal
power capacity and generation. This effect might however be due to the determinist assumptions
used in these runs. Until now, the model included no additional carbon capturing and storing costs,
additional fuel cost clean coal of only 10% above the average fuel cost of coal based generation, a
percentage of COz emissions of 10% of coal emissions, and an initial marginal cost of clean coal
capacity 2006 of €5/W, falling very rapidly after taking-off given the progress ratio of 0.85 and
the assumed cumulative historic clean coal capacity of only 1GW. These assumptions —and the
possible eff and the model as a whole~ will be further explored here.
Anno 2006, potential clean coal power and carbon capturing and storing (CCS) are still charac-
terised by many (fundamental) uncertainties and unknowns and are therefore to be accessed with
broad uncertainty rang mptions and
not certain and different estimates and ranges are advanced by different researchers and part:
Sensitivity analyses therefore seem to be quite necessary. The model sensitivity related to clean
coal power costs could be dealt with in different ways:
are
« Sensitivity related to additional carbon capturing and storing costs:
Additional costs of COs capturing in power plants are estimated to amount to some US$30-
50/tCO, (Moomaw, Moreira, Blok, et al. 2001, p250), 18-70 US$/tCO.
19
mi:
ion 2000, D154), or 15-75 US$/tCO, (IPCC 2005). But these ‘capturing and storing
costs’ also depend on the storing costs which depend on the potential economic application
of the storing of these captured CO» emissions. The total costs might even become 0 or
negative in case of using the CO» emissions for enhanced oil recovery or enhanced coal bed
methane recovery.
Until now, these additional carbon capturing and storing costs in the model amount to
€0/tCOo. Here, amounts of €30/tCO2 and €60/tCO» will be tested.
These costs per tCO» captured need to be transformed to coal based electricity generation
costs in order to be used in the model. Estimated additional carbon capturing and storing
costs per tCO» of €30/tCOz in case of carbon based generation with a CO»-intensity of about
800tCO2/GWh gives additional carbon capturing and storing costs of about €24000/GWh
and about €48000/GWh for a cost of about €60/tCO>.
With the additional carbon capturing and storing costs of €24000/GWh- clean coal power
does not take off in the POLI policy set compared to the very s ‘ul take-off without
that clean coal power takes off very late (after 2070) in the POL2
policy set and that it takes off with much lower fractions in case of the POL3 policy set.
POLI and POL? lead —with this additional cost~ to higher fractions of gas, biomass and wind
(and of renewables in general), to lower fractions of CO» poor generation, to much higher
annual emissions and cumulative emissions, to somewhat higher prices (about €5000/GWh
higher), to lower total cumulative private investments but higher amounts of total cumulative
new capacity installed (more gas).
These evolutions are strengthened in case of additional carbon capturing and storing costs
of €48000/GWh. Then only POL3 takes off but very slowly and only really in the second
half of the century round about the time POL2 would take off in case of additional carbon
capturing and storing costs of €24000/GWh and POLI in case of additional carbon capturing
and storing costs of €12000/GWh.
Sensitivity related to the additional fuel cost clean coal:
Carbon capturing also requires extra energy. These extra energy requirements are estimated
to lie between 15 and 21% (International Energy Agency 2003) p415-420) or between 14 and
25% (IPCC 2005) for IGCC plants and up to 40% for other coal fired plants. These initial
extra fuel needs in the model were initially taken to be 10%. An additional fuel cost of 40%
somewhat slows the penetration of clean coal power. It also slightly influences the amount of
wind power capacity installed, the total and cumulative CO» emissions and other variables
in the model, but does not lead to fundamental behavioural changes.
~ the cumulative historic clean coal
'y and the initial marginal cost of clean
Sensitivity related to the learning curve paramete
capacity, the progress ratio of clean coal power capac
coal capacity:
These three interrelated parameters strongly influence the development of clean coal power
in the model. The impact of changing the cumulative historic capacity of clean coal from
OGW (or actually 1GW what it was until now) té2 10GW or td! 50GW, each time about
halving the fraction of clean coal power. And halving the fraction of clean coal power also
seriously impacts ~especially in the POLI and POL2 runs~ the fraction of gas (which it is
substituted by the relative and cumulative CO» emissions and to a lesser extent the wind
power capacity installed and the electri . The underlying reason for the halving of
the fraction of clean coal power is the less radical decrease of the marginal investment cost
pacity. Now, a less radical decrease of the marginal investment costs might
em reasonable given the maturity of existing coal technologies.
of clean coal
actually s
°POLIsensCCShistcap10, POL2sensCCShistcap10 and POL:
1POLIsensCCShistcap50, POL2sensCCShistcap50 and POL3s
nsCCShistcap10
sCCShisteap50
20
Until now, the initial marginal cost of clean coal capacity was set to €5/W, which is rather
high when compared to estimates in the literature (International Energy Agency 2003) p415-
of about €3/W might be more
appropriate in combination with the cumulative historic capacity of 10GW and the progress
rate of 0.85. This leads ~depending on the evolution- to prices for new clean coal capacity of
of conventional coal power by the year 2015 and of about
the same investment cost (as current conventional investment costs) by the year 2060, and it
also leads to about the same behaviours as discussed before —although the numerical values
e values lead also in the BC to a take-off
of clean coal power. The underlying reason is of course the specific capacity assignment
structure combined with the less prohibitive initial marginal investment costs.
420). A lower initial marginal cost of clean coal capa
about double the investment ¢
slightly differ— with one important exception: thes
selected here because (i) their analyses
are so uncertain that they need to be
But it is not really important which preci
lead to increased understanding and (ii) these variable:
uncertain too. One way to do that is to subject them to multivariate
s done on page[2T|and following pages.
treated a sitivity
analy:
Sensitivity related to the breakthrough of clean coal power:
sensitivity analysis, is the analysis of the consequences of a clean coal
non-occurrence. This could be simulated by assigning a prohibitively high initial marginal
cost to clean coal capacity and removing the initial amounts supported by governments.
Another interesting analysis would be the exclusion of clean coal power from the TECs
scheme.
Related to the previou
Sensitivity Related to the Flexibility Premium: The model is behaviourally sensitive to
different flexibility premia for the flexible gas-based electricity generation: gas-based electricity
generation is not particularly attractive without flexibility premium (BCflexgas00) because with-
out this premium, it is more expensive than coal-based generation. With a flexibility premium
of 10% and without higher gas prices, it is relatively more interesting than coal-based generation
until about 2040 after which coal-based generation becomes more interesting. With a flexibility
premium of 20% and without higher gas prices, it is always more interesting than coal genera-
tion. But higher gas prices (rising to 150% of the BC gas price between 2020 and 2080) undo the
additional advantage caused by the flexibility premia of 10% or 20%. The flexibility premia do
not only impact gas and coal-based generation, but also the wind power capacity installed and
wind power electricity gencrated, and the emissions of CO. They also point at the importance of
(expected) prices and additional stics such as flexibility.
Sensitivity Related to the Pricing Structure: The model contains many ‘simplistic’ struc-
tures among which the elec ing structure, the gencration assignment structure, the new
capacity assignment structure, the electricity and environmental markets structures, et cetera. All
of these structures are open to criticisms and could —if needed- be further tested, elaborated and
improved. A sensible extension of the model would be the addition of feedback loop structures to
render the TE Xs schemes and prices truly endogenous.
The simplis 1
market price plus 1/3 the aposteriori electricity market price plus 1/3 the aposteriori electricity
market price of the previous period. An alternative electricity pricing structure has been thor-
oughly tested: 0 * apriori electricity market price + 1 * aposteriori electricity market price + 0 *
bed above
have been simulated with this alternative structure too. The structure is even more simplistic,
but less robust, and leads to many slightly different numerical outcomes, but not to very different
general behaviours, and some additional insights. These results are not presented here given the
lack of space.
aposteriori electricity market price of the previous period. Almost all simulations des
21
Other Possible Sensitivity Analyses and Alternative Structures: Many other structures
and values are and could be subjected to such univariate sensitivity analyses. Many of them will
now be explored all together.
3.5 Multivariate Sensitivity Analyses
The system dynamics model dis
assumptions (structures, functions and values) and simplifications. Until now, the assumptions
mulated and explored one at a time in order to thoroughly understand the
sed here contains many (sometimes fundamentally) uncertain
have been changed, s
link between these separate assumptions and their resulting time evolutionary behaviours. Now,
a different approach will be followed to grasp the possible joint implications of these uncertain
es.
values, variables and structur
tivity analyses (or Monte-Carlo analy:
randomly selecting values for the uncerte
ised to underly the uncertain phenomena. Multivariate sensitivity analysis
in at least four respects. It could be used to:
Here, multivariat will be performed, simulating
the model 2000 time
distributions hypoth
(MSA) is useful here
in assumptions from uncertainty
s the general behaviour of the policy s
ame time to many uncertain assumption:
© as s (in the particular model) when subjected at
the
compare the resulting behaviour with the behaviours of the separate simulation runs dis-
cussed before and to learn from this comparison;
assess the sensitivity of the model to these changing assumptions;
assess the sensitivity and robustness of the particular policy sets in this model.
The probability ranges/distributions assumed: The following distributions/ranges are as-
sumed in the multivariate sensitivity analyses discussed here:
¢ Initial maximum potentials:
— maximum biomass power capacity potential = RANDOM UNIFORM(75,200)
— initial maximum potential PV capacity = RANDOM UNIFORM(100,2000)
— maximum wind power potential lookup factor = RANDOM UNIFORM(0.8,2)
y = RANDOM UNIFORM(200, 300)
— maximum potential hydro capac
¢ Initial marginal investment costs:
al marginal cost of clean coal capacity 2006 = RANDOM UNIFORM(L.5 10°,4 10°)
al marginal cost of coal capacity 2006 = RANDOM UNIFORM(0.9 10°,1.3 10°)
— initial marginal cost of gas capacity 2006 = RANDOM UNIFORM(0.4 10°,0.6 10°)
— initial marginal cost of wind power capacity 2006 = RANDOM UNIFORM(0.7 10°,1.1
10°)
— initial marginal cost of nuclear power capacity 2006 = RANDOM UNIFORM(L.5 10°,
2.5 10°)
— initial marginal cost of biomass capacity 2006 = RANDOM UNIFORM(10°, 2.5 10°)
¢ Initial average fuel costs:
~ initial average fuel cost biomass = RANDOM UNIFORM(12000,18000)
~ initial average fuel cost coal = RANDOM UNIFORM(5000,9000)
— initial average fuel cost gas = RANDOM UNIFORM(12000,20000)
— percentage CO, emissions clean coal = RANDOM UNIFORM(0.08,0.2)
22
« Progress ratios:
— progress ratio of clean coal power capacity = RANDOM UNIFORM(0.8,0.9)
— progress ratio of biomass power capacity = RANDOM UNIFORM(0.8,0.9)
— progress ratio of coal power capacity = RANDOM UNIFORM(0.80,0.95)
— progress ratio of gas power capacity = RANDOM UNIFORM(0.8,0.9)
s ratio of PV power capacity = RANDOM UNIFORM(0.7,0.9)
atio of wind power capacity = RANDOM UNIFORM(0.8,0.9)
e Other costs:
— additional carbon capturing and storing costs = RANDOM TRIANGULAR(0,48000,0,24000,48000)
— additional fuel cost clean coal = RANDOM UNIFORM(0.1,0.5)
— siting and other investment costs lookup factor = RANDOM UNIFORM(0.5,2)
— maximal additional siting costs = RANDOM UNIFORM(0.1 10°,1.3 10°)
Other parameters and events:
initial decommissioned clean coal power capacity = RANDOM UNIFORM(1,50)
— public support wind generation = RANDOM UNIFORM(0.9,1.2)
~ additional percentage coal use clean coal generation = RANDOM UNIFORM(0.1,0.4)
— flexibility premium = RANDOM UNIFORM(I,1.2)
— year real hydrogen or storage breakthrough = RANDOM UNIFORM(2015,2106)
— percentage CO» emissions clean coal = RANDOM UNIFORM(0.08,0.2)
— electricity intensity EU growth = RANDOM UNIFORM(0.6, 0.66)
— additional growth rate electr
DOM UNIFORM(0, 0.01)
y demand by hydrogen breakthrough constant = RAN-
The premature conversion of idle coal capi capacity of 10% of the coal
capacity idle for over 2 years in a row, kick-starting biomass generation capacity and generation has
been abandoned in this MSA: so there is no premature conversion of idle coal capacity to biomass
capacity. This leads of course to lower fractions of biomass generation, but also to higher fractions
s based and clean coal based generation, to lower fractions of CO»-poor and renewable
generation, and to higher annual and cumulative CO» emissions. But the wind power capacity
installed and annual wind power generation only differ slightly from the high biomass cases. The
results of these analyses will be discussed in the following subsection as part of the multidimensional
evaluation of several policy sets.
y to new biome
3.6 Multidimensional Evaluation of Policy Sets
The goal of this multi-dimensional evaluation is to increase the understanding about the possible
multi-dimensional impacts of strategies, the influence of different preference sets and the sensitivity
of policy choices, not necessarily to determine the best or most appropriate strategy.
The continuous behaviour of the strategies were first of all qualitatively explored on several
dimensions in order to learn from the qualitative behaviour and to eliminate unacceptable strate-
gies. After that, a subset of criteria and specific moments in time wa and an MCDA
method (the PROMETHEE I-GAIA method) was applied to the remaining strategies in order to
gain some understanding about the impact of evaluations on criteria and preferences, and their in-
and the results are not discussed
in this paper.
23
3.6.1 Dimensions, Aspects and Criteria
There are many important dimensions when dealing with the interrelated issue of climate change
and sustainable electricity systems such as the environmental dimension (ENV), the social-cultural
dimension (SOC), the economic dimension (ECO), the technological-technical dimension (TECH)
and the security-reliability of supply dimension (SEC).
There are also important criteria within these dimensions —many of them repres
ables in the system dynamics model- that could be looked at. The following seven specific cri-
teria/proxies will be focussed on here: (1) fi: percentage CO2 emissions base 2006 [%] (ENV):
(2) fo: fraction of CO2 poor generation of total generation [%| (ENV/TECH); (3) fs: fraction
of renewable generation of total generation [%] (ENV/TECH); (4) fa: fraction of gas based gen-
eration of total generation [%] (proxy for SEC); (5) fs: electricity price [E/GWh] (SOC/ECO);
(6) fe: total private cumulative investment cost 2006-... [€] (ECO); (7) fz: wind power capacity
installed [GW] (ECO/TECH/SECTORAL).
ented by vari-
3.6.2 Strategies
Many strategies could be simulated with the model. The multi-dimensional evaluation phase is
illustrated by means of seven relatively simple strategies, all based on the policy sets discussed
before.
The strategies are subjected to the multidimensional sensitivity analysi
[B.5|which is characterised by low biomass capacity and no premature conversion of coal to biomass
based generation capacity and which leads to results similar to those found in the literature. The
initial wind power capacity installed is taken to be 40.504GW and the initial new wind power
capacity under construction 6.183GW /y. The reason for using these very precise numbers is that
the results will most probably be interpreted by some in a numerical sense, no matter how many
warnings against such interpretations.
Strategy So is simply the BC policy set, strategy $, the POL1 policy set, Sz the POL2 policy
set and $3 the POL3 policy set, all subjected to these new (initial) conditions. Strategy Si
corresponds to the POL2 with a sudden new politically forced new nuclear capacity commissioning
of 20GW in 2015, 50GW in 2020, 50GW in 2030, and 50GW in 2050. Strategy $5 corresponds to
POLS but without clean coal. And strategy Sg corresponds to POL3 with wind subsidies of 30%
until 2050 and a real hydrogen breakthrough between 2015 and 2025.
discussed in subsection
3.6.3 Multi-dimensional MSA Dynamics
The resulting evolutions of the IQRs (Inter Quartile Ranges) and 80% IPRs (Inter Percentile
Ranges) of the strategies on these criteria are displayed in table [1] (124). The table contains
an cnormous amount of conflicting information that would be lost in aggregated form or without
proper analysis and that is difficult to be used for decision-making without additional information.
Examples of such conflicts are (i) performance conflicts of a strategy on different dimensions (for
example f;(So) versus f5(So)), (ii) performance conflicts of a strategy on different moments in
time (for example f7(S¢)), or (iii) performance and evolution conflicts of a strategy on a dimension
and moment in time due to uncertainty (for example f,(S5) on any moment between 2030 and
2075).
The criteria are also to be interpreted in different ways. Some criteria give an indication of
the absolute performance of strategies on a criterion (f;) whereas other criteria might be used
to evaluate the relative performance of different strategies on a criterion (fs). And some criteria
provide information about the absolute performance of a strategy on a criterion but hide the
relative performance of a strategy on a criterion (f;(S¢) suggests the absolute amount of wind
power capacity installed in case of Sg but does not indicate that the wind power fraction IQR of Ss
tops round 29-41% in 2053). The graphs also give an indication of the in/sensitivity or robustness
of the strategies in the model.
Many qualitative conclusions might be reached based on this table, for example (i) that no
strategy is unambiguously good or bad on all criteria at any moment in time, (ii) that Sp is
(souy (werpour) pue somy penprarput
UO UT UIO0Z 0} TOISIAA OTMOTAIO ot]} 0} Parr ore sIOpeat) UNI TOTY[NUMIS Werpout ay} sytosordor out] YouTG oy} pur ‘%9g vore Aor yy BI ony ‘sums
= UOL}LINULS at] JO YG suIeyUoD vere AoIS Yep oY} :BLIo}LIO Wades UO salSoye.1ys XIs Jo sormeUAp AreuOLN]oAe outr} ot] Jo MOLeNfeAd aATyeyyPENY 27 epqvy,
= = a " ——$<$ | a — Sg
Yin <a : E
: = == : = : =| = - = a5
_— —_— -_ | _— _-—
SS
= = : | = =A = =
CC Oe | sEEEBEGEE
: ~~
= | z : : ar : m | 7. ———= ¥s
_—«< —_— Lae Ea! =
Bane ade
aL : = | = : =e | = —
= _ | —— a —_—_—_—_
se : Shenae” _ “
=| = _ oe . 1S
i a a _—
i ae
= _ = : ee me ae : _—
—_— La —_— a |
(eur) [M0007] = (urer) [> ,,0T] = (urut) : (urar) [%00T] = (xen) [4007] = (xem) taal = aa [%o0s] =
porreysuy Ayoedeo -9002 Sttourysoaty oywatd [4uap/o000023] uorerouas poseq uorperou98 914 uorerottes 100d 900z aseq storsstur9
womod pun :2f 3g sm 2s sorid Apouyo9[9 se8 uoryoey :tf -emouor uorjsey iff Zo jo uonowy + ef QQ eBequooiod :1f
25
extremely bad on f; (the percentage CO. emissions) and several other criteria, (iii) that 53 and
Sg are good on all criteria considered except the direct electricity price, (iv) that $3 and Sg show
that the fraction of gas based generation needs to be reduced in order to effectively reduce CO.
emissions. Such a table could then be used to eliminate unacceptable strategies.
The long term evolutions on criteria fy and fy indicate whether the CO, emissions and CO,
poor fractions are heading in the right direction: $3, Ss, S¢ are heading in the right direction,
Sp and S4 not really, and So and S; absolutely not. Strategies Sy and S; might therefore be
rejected on ethical (intergenerational) grounds. The $2 and $4 policy sets might on the other
hand be strengthened and redirected after 2030 and might therefore not necessarily be rejected.
for other criteria and time increases the understanding and allows to reject other
strategies too. The price criterion might also lead to the climination of other strategies. This
arily needs to be performed by the parties involved: just handing over conclusions
ive. Such tables could then be used as very powerful tools for exploration, critical
reflection and dialogue. They could also set the stage for more formal/quantitative analyses which
are not discussed here.
4 Some Conclusions, Ex-post Criticisms and Venues for Fu-
ture Research
4.1 Conclusions
Even without the possible extensions, valuable lessons for climate change policy-making could be
learned from this model, for example that:
¢ The different long-term climate change policy sets differently impact the dynamics of the
specific emissions of the power mix and the electricity demand, and therefore the CO,
emissions. If a one-time policy set is chosen among the POL1, POL2 and POLS policy sets
taking their MSA distributions into account, then only POL3 leads ~generally speaking~ to
significantly reduced annual CO» emissions and convergent cumulative CO emissions.
e The CO2 emissions are
— only temporarily sensitive to abrupt nuclear phase-outs, but are sensitive to abrupt
governmental nuclear initiati
sitive to high gas prices in case of moderate climate change policies
— slightly dependent on a hydrogen/storage breakthrough,
— very dependent on a hydrogen/storage breakthrough accompanied by an increased elec-
tricity demand,
— very sensitive to variations in the GDP growth rate,
sensitive to the initial maximum wind power potential assumption in case of a hydrogen
breakthrough.
Valuable lessons for boosting wind power capacity installed could be learned from this model
-this could be done for other gencration technologies too, but is not explicitly done here~ for
example that:
© The successful (very) long term development of wind power strongly depends on several
related pol and/or evolutions, such as strong green/CO2-poor investment/generation
policies, a ful breakthrough of cheap hydrogen/storage technologies, the low poten-
tial and little su of other especially of other CO2-poor and/or intermittent— generation
technologies, (expected) long term fossil fuel prices, continuously increasing electricity de-
mand, and so on.
26
« In this model, the wind power capacity installed (and consequently the wind power gener-
ated) seems to be
— marginally sensitive to varying siting costs,
— slightly sensitive to high gas prices without a hydrogen/storage breakthrough,
— very (behaviourally) s
additional demand),
sitive to a hydrogen/storage breakthrough (with and without
— sensitive to varying maximum wind power and other renewable potentials,
— very sensitive to varying GDP growth rates,
ive to a doubling of the maximum wind power potential without a hydrogen/storage
breakthrough, and very sensitive to a doubling of the maximum wind power potential
with a hydrogen/storage breakthrough,
— sensitive to very sensitive to variations in the maximum potentials of other maximum
renewable potentials,
— very sensitive —but only in the long run and in case of POLI and POL2- to variations
in carbon capturing and storing cos!
— and extremely sensitive to the in/activation of the structural intermittence assumption
binding wind and PV power.
e A sustained and large-scale wind power breakthrough therefore requires a breakthrough by
2020 of a cheap storage/hydrogen technology and/or backup generation technology and/or
other technologies —and thus much research into these related technologies~ to solve the
intermittence/variability problem.
« A temporarily subsidised wind power boom might without such a storage/hydrogen tech-
nology breakthrough- end in an overshoot and subsequent collapse of EU-25 wind power.
Long term perspectives, policies, goal setting and focussed research are therefore necessary.
¢ Wind power might face a bright future in the EU-25 if such a breakthrough occurs or if
PV power and wind power are not linked in terms of their intermittence. However, most
simulation runs do not lead to very high fractions of wind power generation.
« Projections of wind power capacity installed vary strongly: comparing the IQRs of POLI,
POL2 and POLS, it could be said that POL3 is best in terms of wind power capacity installed
by 2010, that POL2 is best between 2020 and 2030, and POLI scems to be best afterwards,
mainly caused by the higher electricity demand.
e Polici
dec
s to boost wind power capacity installed do not necessarily lead to significant long-term
s of CO» emissions, and vice versa. The absolute amount of wind power capacity
installed is -in the long term- lower in case of strong climate change policies wh
that tringent climate change policies might in the long term be more interesting for
the wind power industry. The goal of maximising wind power penetration is therefore not
sarily the same as that of reducing CO» emissions and concentrations. Persistent growth
of electricity demand might for example help the development of renewables in absolute
terms, but is detrimental in terms of CO emissions.
means
nec
© One should be careful with policies to boost one type of technology ~in casu wind power~
which might seem attractive from a financial point of view, because ’[...] adding technologies
to the portfolio will increase the need for learning investments, and if the added technologies
compete for the same learning opportunities, they will delay break-even and reduce the
present value of the portfolio. An efficient portfolio must balance allocation of learning
opportunities against the need to diversify energy supply and to spread technology risk’
(International Energy Agency 2000) p83).
27
Some of the many possible less
generation are that:
ms to be learned from this model concerning fossil fuel based
gas might substitute for coal, and as such slow the introduction of renewables over time,
the further evolution of gas b
the (expected) gas market. pri
sed clectricity generation is very (behaviourally) sensitive to
(expectations of future) gas pric
mix;
strongly influence the development of the future generation
and clean coal generation might become ~in case of sufficient start-up investments, rela-
tively low costs and inclusion in the TECs schemes~ a very strong competitor to renewable
technologies.
Some conclusions related to the needed investment costs are that:
« cumulative investment cost might be significantly higher than suggested in the literature and
slightly higher than industrial forecasts. The IQRs of the three policy sets suggest that the
cumulative investments in new generation capacity from 2006 on might amount to €100-200
billion in 2010, €500-750 in 2020, and €1000-1500 in 2030, which is slightly higher than
industrial forecasts
« the amounts of investments needed are lower in case of more sustainable electricity systems
with lower electricity demands;
investments in electricity infrastructure are necessarily cyclic
ven with continuous capacity
additions which is not even the case in reality),
the cumulative inves
climate change polic
tments are very sensitive to GDP growth and —in case of moderate
rather sensitive to (high) gas pric
it determines
to inves
the choice of generation technolog' in is extremely important becau:
which technologies will (most likely) be used for future generation and which technologies
will become ever cheaper and therefore ever more competitive.
And some ~among many~ other possible conclusions are that:
simulated
electricity prices are proportional to the stringency of the climate change polic
if climate change damage and adaptation costs and other external costs are not included as
in this particular form of the model,
electricity prices are when taking the whole picture (possible material and fuel price rises
climate change damage costs, needed investment costs, potential excess demand, et cetera)
into account~ not likely to decrease much ~in spite of the efficiency gains and technological
development~ in the medium, long and very long term,
apparently insignificant decisions/events might lead to very different evolutions over time.
However, the fundamental direction of these markets is not a question left to be answered
od in this paper~ might help to make
ple consequences of such insignificant
by 'the market’ alone. Models —such as the ones discuss
ions based on a better understanding of the po:
decisions/events,
de
« there are “free hineh
place.
if gradual (but rather drastic) electricity demand reductions also take
28
s model has been discuss
of several parameter
A particular structural form of a system dynamic od here, followed by
and in case of two
contains many simplifications such as the treatment of
nuclear power generation as must-run generation, the lack of competition between PV and other
generation technologies, the aggregation of all wind power technologies or all hydro technologies,
et cetera.
Some possibly interesting extensions and improvements that could be explored extending this
model are among else:
© ‘economy-energy-environment-climate change’ loops for real long-term policy and integrated
cost analyses, for example the necessary feedback loops to render the TECs and TGCs
emes truly endogenous;
« the interaction with other energy markets, such as the heating market, the mobility market,
or the gas market;
e structures to explore revolutionary technological breakthrough:
logical evolution by means of temporary steeper learning curves
and/or accelerated techno-
additional modules/structures dealing with other technologies for example oil based gener-
ation, fuel cells, CHP, et cetera, or more detailed (sub)technologies/technological differenti-
ach as onshore and offshore wind power instead of just wind power;
ations s
subscripts dealing with the EU-25 member states in order to explore the issues of the EU-25
“energy islands’ and their interconnections;
detailed structure:
structure used here;
replacing really simplified structures such as the simplistic siting cost
new endogenous structures to deal with social and institutional aspects related to wind power
penetration, such as the expected public support for wind generation, more or less favourable
ruling and favourable spatial planning policies, increasing/decreasing delays, et cetera;
detailed structures to deal separately with the EU-25 and the world component of learning
curve effec
more elaborated structures to differentiate between expectations and ‘real’ evolutions.
4.2 Ex-post Criticisms and Venues for Future Research
Different system dynamics models or extensions of this system dynamics model should/could
be developed to explore th ues. The potentially important influences not
implicitly and explicitly modelled matter because of the fact that these omissions are among the
most important modelling assumptions made.
Potentially interesting related is and aspects not dealt with here
are for example: (i) the degree of competition between energy companies, their bounded ratio-
nal behaviour and stre (ii) the precise structures of these energy companies, their
po solidations; (iii) regional and national boundaries and borders (although
still very important) and detailed country-specific electricity trading and support structures; (iv)
aspects of financing; (v) the broader/world demographic evolution, the broader energy domain,
economy, climate change; (vi) international fuel supply, demand and availability; (vii) technical
aspects of clectricity grids; (viii) the distinction between peak load and base load; (ix) specific
(sub)technologies such as CHP, et cetera.
The purpose of the model was the exploration of the development of electricity generation
technologies -more specifically wind power generation technologies~ in order to increase the un-
derstanding of its multi-dimensional dynamic complexity. Given the focus on analysing the compe-
tition between developing technologies, one might say the primary decomposition has been along
related or other is
29
the lines of technology-centered subsystems. Aspects of the issue explicitly dealt with were there-
fore the dynamic complexity aspect and the multi-dimensionality aspect, and to some extent the
uncertainty aspect related to uncertain parameters and exogenous variables. Important aspects
of this issue that were not dealt with include among else (i) the multi-actor aspect (divergent
emotions of the many stakeholders and actors involved), (ii) the multi-level
aspect (the many policy/physical/.. ) (iii) the uncertainty aspect in its full
depth ard -width, (iv)-the network aspects: (both in’texma of physical networks euch as electricity
grids, as in terms of actor, technology, energy or financial networks), (v) the geographical aspect,
et cetera.
The ex-post use of conceptual frameworks helped to put the model
and results into perspective too. Using a conceptual level framework, it could first of all be argued
that it would have been better to model also at the level of the individual member states since that
is currently the relevant level of policy-making. Second, it could be used to explore the included
and omitted exogenous variables (inputs for this model) and output variables (inputs for other
systems or models) which make up the interface with other (sub)systems and (sub)models. The
influence of all exogenous variables should be thoroughly explored. This exploration in the case of
the system dynamics model shows that some higher-level and lower-level systems should actually
also be modelled because they deliver crucial dynamic inputs to this model that are not readily
available in the literature or as outputs of other models.
Third, interactions with other related subsystems are not explored because of the fact that the
necessary models are not compatible (in terms of purpose, time scales used, etc.) with the sy
dynamics model under consideration. Examples of omissions because of method-misfit are detailed
|-models. This leads to the implicit assumption that there is only one perfect grid, or multiple
interconnections and infinite capacity. The influence of limited, constrained and
C4 should therefore be ass d and —if important be included in
some way in the system dynamics model. The outcomes will then most probably be worse than
suggested by the current model. This also shows that methods and models used determine what
could actually be explored and what possible outcomes could be. And policy recommendations
are in turn biased by the methods and models used through these outcomes.
Fourth, combining the aforementioned System-of-Systems hierarchy level view with the hor-
izontal energy sector view (as in (Agusdinata 2006)) shows that the model boundary
narrow: only the competition between these 9 generation technologies is
elled. Many other influences are nevertheless modelled semi-endogenously s
surprises, world fuel pr regional and local policies, and even the societal
value system. But still, it shows the necessarily limited exploration of plausible influences such
as the influence of currently unknown emerging technologies which are of course fundamentally
uncertain.
and availabiliti
References
Agusdinata, D. (2006, April). Specification of system of systems for policymaking in the energy
sector. In Proc. of the 2006 IEEE SMC International Conference on System of Systems
Engineering Los Angeles, pp. 211-217. [29]
AMPERE Commission (2000). The AMPERE report. MinEco. (19)
Ericsson, K. and L. Nilsson (2006). Assessment of the potential biomass supply in Europe using
a resource-focussed approach. Biomass and Bioenergy 30, 1-15. [17]
). An energy policy for Europe. Number SEC(2007) 12
European Commission (2007, Jamua
COM/2007/0001 final. [1]
European Energy Agency (2005). How much biomass can Europe use without harming the
environment? EEA Briefing 2. {I7|
Forrester, J. (1994). System dynamics, systems thinking, and soft OR. System Dynamics Re-
view 10(2-3), 245-256. 9]
30
International Energy Agency (2000). Experience curves for energy technology policy.
IEA/OECD. [26]
International Energy Agency (2003). World Energy Investment Outlook. 2003 Insights. Paris:
OECD/IEA. {19} 20)
IPCC (2005). IPCC special report on Carbon Dioxide Capture and Storage. Prepared by work-
ing group III of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge
University Press. 442p. [19]
Moomaw, W., J. Moreira, K. Blok, et al. (2001). Climate Change 2001: Mitigation, Chapter
Technological and Economic Potential of Greenhouse Gas Emissions Reduction, pp. 167-277.
Cambridge and New York: Cambridge University Press. [I8)
Parikka, M. (2004). Global biomass fuel resources. Biomass and Bioenergy 27, 613-620. [17)
Pruyt, E. (2007a). Dealing with uncertainties? combining system dynamics with multiple crite-
ria decision analysis or with exploratory modelling. In Proceedings of the 25th International
Conference of the System Dynamics Society, Boston, MA. em Dynamics Society. [4]
Pruyt, E. (2007b, January). Decision-Making and Dynamically Complex Multi-Dimensional
Societal Issues: Combining System Dynamics and Multiple Criteria Decision Analysis to
Explore the Energy-Climate Change Issue. PhD Thesis, Vrije Universiteit Brussel, Solvay
Business School, Brussels
Pruyt, E. and W. Thissen (2007, April). Transition of the european electricity system and
system of systems concepts. In Proceedings of the 2007 IEEE SMC International Conference
on System of Systems Engineering, pp. 1-6.
Appendix — The EU-25 Power Sector: a System
Dynamics Model of Competing Electricity
Generation Technologies
Erik Pruyt
Delft University of Technology
Faculty of Technology, Policy and Management, Policy Analysis Section
P.O. Box 5015, 2600 GA Delft, The Netherlands
Tel: +31/152787468 — E-mail address: E.Pruyt@tudelft.nl
August 22, 2007
1 Feedback Loops — Feedback Loop Diagrams
The feedback loop diagram in figure[I]is broken down in feedback loops concerning wind power
(la, 2, 3, da, 5a, 6a, 7a, 8a, 9a, 10a, Lal, 1Ha2, 12a, 1a) and feedback loops concerning the other
generation technologies considered [technology i] (1b, 4b, 5b, 6b, 7b, 8b, 10b, 11b, 12b, 13 and
14b). See figure [Jj for a clearer picture of feedback loops la, 2, 3, 4a, 5a, 6a, 7a, 10a, Hal and
11a2. Sce figure{l|for feedback loops 8a, 8b, 9, 12a, 12b, 13, 14 and 15. The other electr
ation technologies (technology i) considered are gas-based, coal-based, nuclear-based, (potential)
clean coal-based, biomass-based, solar PV, hydro and geotherm electricity gencration technologies.
When the corresponding feedback loops are similar ~such as feedback loops 1a and 1b~ then only
one feedback loop will be discussed, mostly the one concerning wind power. This break-up makes
the feedback loop diagram more complicated than strictly necessary, but also allows us to focus
on wind power. The following feedback effects are generally speaking~ dealt with:
y gener-
Feedback loops la and 1b — the positive experience-cost loops : Investments in
wind power capacity increase the cumulative historic wind power capacity sold which increases
learning and experience of the wind power sector which decreases the marginal investment cost
of new wind power capi installed in the future which increases the relative attractiveness
of new investments in new wind power capacity, which leads in turn to even more investments
in wind power capacity, and so on, or in terms of feedback loop la: investment wind power
wind power
marginal investment cost new wind power capacity exp margin new wind capacity ¥ relative
attractiveness investment in new wind power capacity and again~investment wind power capacity.
The investment costs of the other technologies considered also decrease with experience gained,
following their specific learning curves (see feedback loop 1b).
capacity ~ cumulative historic wind power capacity sold ¥ learning and experience
Feedback loop 2 — the positive wind power potential expansion loop © : The decreasing
marginal investment costs of new wind power capacity installed increase the maximum wind
power potentiality, increasing the gap with the wind power capacity installed (see feedback loop
3), decreasing the siting costs (more sites available given the lower investment and other
investment costs, increasing the relative attractiveness of investing in new wind power capac
which —ceteris paribus- leads to more investments in wind power capacity, to more cumulative
historic wind power capacity sold which increases the learning and experience of the European wind
leaming and experience
wind power
¢ of cx tition in
sujal’ inccaions-coaioon baat
‘investment cost
ao ~”~C Rewind mascum percentage growth + relative attractivene:
max wind power =” powercapacity rind power construction investment cepeity
tential { =| technologyi> 9
2 potential 6 Po
specie at electricity cumulative
Ga expansion loop price inSy> | historic wind
_/ exp margin new power
additional
= future decon
stalled capacity ot
~ yt
+ vd cpenty capa capacity sold
<> siting and other 79. + ‘
gap windpower - investment | \
costs relative attractiveness 2
Potential 5 maximum investment in new wind investment vind $$ expected new:
Saree 109) power capacity power capacity capacity to be
Ay : installed indy
<wind power ercentage varible 12a.backup and X 509 new capacity Loop
capacity. etaled> output capacity om Ad
: mm m
public acceptance of profit A ity R_,5 Ks ae
wind power a -generatio: * power loop Ad
_ aptiori generation installed po capmetty Sainte decommissioning
cost wind power # ; : Beier)
5a competition for sei oe power capacityy
4 A\I generation Loop 591 attr genby shetty
cost wind power Ay ‘wind power potentially
show sold a Heleoaietcy experienc:
“nese
attrectiveness flexble and “wnt powerchetecy +
storable backup generation t eet and: on
‘and storage capacity
a ak df ctr + tikcaina, jerience-
* SOF "mat *
_ TGCs and a. wind power
‘TOGs and TECs price * race supply Pe
+ ene!
expected future TGCs ag ~~ more espected oneienay
wal CO2 TECs prize ‘ wr demand insy.
+ +
electricity de
expected Paton renal me = effectiveness
price inSy DSM and REU
profitsbiity ost electricity TGCsand “#4 TOCs and TECs
eer Spe generated by) TECs demand + required bs
“A technology 1 CO2 emissions greenness
- G a4 ao - + ‘by! ot ‘societal
aS i eration
ment c en cael SP compeition for Schmologes value
rarginal investment eration generation loop
costs other generation y “aah technology i Gh ohctilty ponantnd 2 —
technologies _elative attractiveness.ag— ‘0°! yr OO cat cold by eineaatiin
investment capacity . PMP ‘‘ vacasine ye
i b AD wehbe rl wrk ses capacity technology i
an es re ara = specific CO2 emissions
fuel dependence nology: erated technology i
ears aad ‘aiveatliend Sabtinily slecticity prorat icing” dion Cool TBE S ley
fs |, technology i generation of total
saumenatnve 12fuelimport future decommissioning electri
ning, ‘ity potentially
capacity sold “ curaulative learning and
installed capacity other generated by generation
technology i panto Done cxpecty technology i experience technology i
=relati
parate installed in y> HR eared
installed technology i
Figure 1: General feedback loop diagram of the EU-25 wind power model
learning and experience
wind power
marginal investmeit cost
, ie new wind power capacity
max wind power
potential ‘ cA
potential - _ Lexperience-cost loop
SR a peniton lien i
cumulative historic wind
exp margin new
% wind capacity
‘“Sxpected electricity
sap ting ie, Sitingand other : + cs> demand inSy
“power Savormsenbooets ‘relative attractiveness
potential 3 maximum investment in new wind"? investment wind". "
rer Ca bs wer capacity Se re ae
ease pala aie to be installed inSy
= aelaeaal a
cost wind power ryind power Y penal se ghana
electricity sold pe
wind power electricity >
a 2 fn
slsctrivity price generated and sold wind pow
pa kin capacity factor
Zadiminishing by ov
eft Leep 4D
TGCs and
TGCs and TECs price — _—- TECs supply
Figure 2: Feedback loops la, 2, 3, 4a, 5a, 6a, 7a, 10a, 1lal and Ma
power sector which decreases even more the marginal investment cost of new wind power capacity
installed, and so on, or in terms of feedback loop 2: marginal investment cost new wind power
capacity~max wind power potential ¥ gap wind power potential~siting and other investment costs
relative attractiveness investment in new wind power capacity ¥ investment wind power capacity
F cumulative historic wind power capacity sold ¥ learning and experience wind power~marginal
investment cost new wind power capacity.
Feedback loop 3 — the negative maximum potential loop ~ : The more wind power
capacity installed, the smaller —ceteris paribus— the gap to the maximum wind power potential
becomes, which makes investing in new wind power less interesting, especially because of increasing
siting and other cost estments in wind power capacity and hence —with a
delay— also the increase in wind power capacity installed, and so on, or in terms of feedback loop
3: wind power capacity seatallellsgap wind power potentialsiting and other investment. cos
relative attractiveness investment in new wind power capacity ~ investment wind power capacity
~ wind power capacity installed.
s, which slows new inv
s>
Feedback loops 4a and 4b — the positive profitability loops @ : The name of these
positive feedback loops and one of their causal links is displayed in grey in the feedback loop
diagram for two reasons: (i) because it contains a clear simplification of reality, and (i
of its relatively modest influence on the model. There is namely no real financing and im
module in the model which would allow the use of own capital and available cash —both generated
by past profits~ for additional capital investments. Therefore the assumption
profitability of capacity inv
the investments in capacity and after a delay also the capacity installed, which in turn increa
the electricity generated by the technology. Now, the more clec
the ins
generated or sold, which increa
because
ment
made that current
lightly influences the relative attractiven ments, which drives
es
icity potentially generated by
alled capacity, the more electricity that could be sold, the lower th
-given the electricity price~ the profitability of the technology,
and so on, or in terms of the wind power feedback loop (4a): profitability wind power ¥ relative
stment wind power capacity @ wind
power capacity installed $ wind power electricity potentially generated ~ wind power elect:
st of the electricity
attractiveness investment in new wind power capacity inv
generated and sold~cost wind power electricity sold=*profitability wind power. Feedback loop 4b
represents the same profitability-investment mechanism applied to the other technologies
Feedback
1 and 5b together represent the competition between the technologies installed to be used
Feedback loops 5a and 5b — the positive competition for generation loops @:
loops
for generation purposes, or in terms of wind power generation (feedback loop 5a): wind power
tricity generated and sold cost wind power electricity sold 2 apriori generation cost wind
power — relative attractiveness generation by wind power ¥ wind power electricity
enerated and
sold. Moreover, the apriori generation cost of technology i in feedback loop 5b increases ~in case of
generation dependent on (fo: ) fuels if the (exogenous) fuel price of the fuel required
increases. The fuel prices are not modelled endogenously, because the endogenous fuel demand is
ss, fuel prices and even
will most probably seriously influence the wind energy dynamics.
enarios are considered and the sensitivity to different fuel pri
, nuclear,
only a small fraction of the total demand on the world markets. Neverthele:
more the expected fuel pric
This is why eral fuel price
are explored later on.
Feedback loops 6a and 6b — the negative decommissioning loops ~ : These negative
feedback loops decrease with some delay the capacity installed through decommissioning of the
TNote that this shadow’ variable is displayed as <wind power capacity installed> in the feedback loop diagram|
it already appears elsewh
causal link is computed slightly differently in the simulation model. There, the a priori costs are calculated
by means of the new information about the capital invested, but with for example the information about the TGC
prices of the previous year.
‘e in the model.
pacity installed at the end of its lifetime, or in terms of feedback loop 6a: wind power capacity
installed ¥ future decommissioning installed wind power capacity? wind power capacity installed.
Different decommissioning model structur ioning after exactly the lifetime,
or the annual decommissioning of the capacity installed by 1 over the lifetime~ are possible and in-
fluenci
such as decommi
the behaviour of the model. This feedback loop, together with positive feedback loops 10a
s of capacity
arily the same technology as the decommis-
and 10b leads to replacement investments, proportional to the relative attractivene
investments of the respective technologies
sioned one.
Feedback loops 7a and 7b — the negative diminishing profitability loops : Feedback
loop 4a is weakened in case of a (too) successful development of wind power capacity installed
because of the decreasing market price of the CO tradable emission certificates (TECs) and
tradable green certificates (TGCs) with increasing TGCs and TECs supply given their respective
quantities required and their resulting demands, or in terms of feedback loop 7a: wind power
electricity potentially generated? TGCs and TECs supply TGCs and TECs price (given the
TGCs and TECs demand)~cost wind power electricity sold.
The diminishing profitability loops of other non-renewable technologies (feedback loop 7b) are
characterised by a slightly different mechanism. Their generation costs increase with the cost of
acquiring TGCs4 and/or CO. TECS®.
Feedback loops 8a and 8b — the positive generation cost reduction loops @ : Decreas-
t of
Id, and will -given the electricity price result in a higher profitability of that technol-
ing marginal investment cost of new capacity installed will gradually /slowly decrease the c
electricit
ogy, and so on, or in terms of feedback loop 8a: marginal investment cost new wind power capacity
f cost wind power electricity sold profitability wind power ¥ relative attractiveness investment
in new wind power capacity ¥ investment wind power capacity ¥ cumulative historic wind power
new wind power
haracterised by
capacity sold learning and experience wind power marginal investment co.
capacity. Feedback loop 8b dealing with the other generation technolog:
the same structure.
Feedback loop 9 — the positive capacity expansion of the wind power construction
industry loop @ : The capacity expansion of the wind power construction industry loop is —at
first sight— a positive feedback loop structure which combines push and pull effects of the wind
power ci stments in wind power capacity (pull), the bigger the
capacity of the wind power construction industry will become, and the bigger the wind power
struction industry: the more inv
construction industry, the more push leading to more investments in wind power capacity. This
for example the case in the models discussed in (Pruyt 2004): there the capacity of the wind
power construction industry drives the commissioning of new wind power capac
the current model, the variable marimum growth wind power construction industry only tops the
in wind energy rise by more
also seen in the real-world®, This
. However, in
growth of the wind power construction industry if the investment
than a certain percentage —this limited capacity expansion
negative influence tops and thus slows both capacity expansions and should therefore be seen as
above the maximum growth rate
ables displayed in the feedback loop
a negative feedback loop only if the rate of expansion incre
of the wind power construction industry. In terms of the v:
3 Wind power electricity potentially generated drives the TGCs and TECs supply in the feedback loop diagram
because the electricity generated might not be sold, but still generates TGCs. This is not the case for TECs. In
the stock-flow diagram, the detailed structures are elaborated (more) correctly.
4 Electricity generated and sold by generation capacity technology i + TGCs and TECs demand ¥ TGCs and
TECs price ¥ cost electricity generated by technology i, and so on.
5 Blectricity generated and sold by generation capacity technology i ¥ CO2 emissions by other generation tech-
nologies ¥ TGCs and TECs demand ¥ TGCs and TECs price © cost electricity generated by technology i, and so
on
Currently, production capacity is indeed a limiting or slowing factor for several wind turbine producers and
wind power plant developers such as for example Gamesa —worldwide the second biggest player with a market share
of more than 18%- whose demand exceeds their production capacity even after expansion.
diagram, this gives: investment wind power capacity capacity wind power construction industry
-which is possibly topped by maximum growth wind power construction industry-¥ investment
wind power capacity.
Equivalent loops are considered for other modules dealing with niche market technologies with
enormous growth rates (such as solar photovoltaic (PV) power), not for conventional technologies.
The assumption behind the unlimited growth rates of the capacity of the conventional energy
technologies is that these industries are mature. It might however be possible to vary slightly
the construction delay of these mature construction industries with the respective new capacities
joned. This is not explored here.
commiss
Feedback loop 10a and 10b — the positive new capacity required loops @ : These
positive feedback loops reflect the fact that the more installed capacity is decommissioned —ceteris
paribus~ the more capacity replacements are required to satisfy the electricity demand, or in terms
of feedback loop 10a: investment wind power capacity? wind power capacity installed ~ future
decommissioning installed wind power capacity ~~ expected new capacity to be installed in 5y¥
investment wind power capacity. The forecasting time horizon is taken to be 5 years here —which
is of course a simplification. The expected new capacity to be installed in Sy is also fuelled by the
future decommissioning installed capacity other generation technologies, and by the expected new
capacity to be installed inSy.
And the proportion of this expected new capacity to be installed in 5y 2
technology depends on the relative attractiveness of investing in the particular technology (rela-
tive attractiveness investment in new wind power capacity and relative attractiveness investment
capacity technology i).
signed to a particular
en
Feedback loops 11al and 11a2, and 11b — the positive experience improvement loops
@® and @ : Feedback loops 1lal and 1la2, and 11b are related to feedback loops la and 1b
respectively —which deal with the decreasing investment costs with increasing experience— because
they embody different aspects related to increasing experience and learning. Feedback loop 1lal
deals with the increasing capacity factor with increasing learning and experience: learning and
experience wind power wind power capacity factor ¥ wind power electricity potentially generated,
and so on (see loops 1a and 2). Related to this feedback loop is feedback loop 1a2 concerning
the dynamic gap of wind power potential: gap wind power potential f wind power capacity factor
Y wind power tricity potentially generated, and so on (see also loops la and 2).
And feedback loop 11b is related to loop 1b and reduces the marginal CO» emissions”! with
increasing learning and experience of the respective CO» emitting technologies.
Feedback loops 12a and 12b — the negative backup and storage loops ~and~ : The
negative backup and storage loops 12a and 12b make investments in flexible and storable backup
generation and storage capacity increasingly attractive with increasing percentages of capacity
characterised by variable output (such as wind power capacity), or in terms of feedback loop 12a:
percentage variable output capacity~relative attractiveness investment in new wind power capacity
investment wind power capacity % wind power capacity installed percentage variable output
capacity; and in terms of feedback loop 12b: percentage variable output capacity ¥ attractiveness
flexible and storable backup generation and storage capacity ¥ relative attractiveness investment
capacity technology i¥ generation capacity installed technology i= percentage variable output
capacity.
7Feedback loop 1b: learning and experience technology i=* specific CO2 emissions generated technology i +
C02 emissions by other generation technologies, and so on, back to learning and experience technology i, and so
on.
8Only the attractiveness of flexible and storable backup generation and storage capacity is increased, not the
attractiveness of technologies characterised by low degrees of flexibility such as nuclear energy.
Feedback loop 13 — the negative fuel import dependence loop ( : This feedback loop
only applies to those technologies burning fuels not produced in the EU and makes new capacity
investments of these technologies relatively less attractive the more fuel is required by existing
plants, or in terms of feedback loop 13: electricity generated and sold by generation capacity
technology i ¥ fraction fuel i based generation of total ~ fuel dependence y generation
relative attractiveness investment capacity technology i generation capacity installed technology
iy electricity potentially generated by generation capacity technology i ¥ electricity generated and
sold by generation capacity technology i.
Feedback loops 14a and 14b — the negative expected TGC and TEC cost loops ©
and : Feedback loops 14a and 14b —which are strongly related to feedback loops 7a and 7b—
are represented in a simplified way in the feedback loop diagram. There it is indicated that
the TGC and TEC prices also influence the expectations about the future TGC and CO, TEC
price and thus the relative attractiveness of new capacity investments, or: TGCs and TECs price
+ expected future TGCs and CO TECs price ¥ relative attractiveness investment in new wind
power capacity, and relative attractiveness investment capacity technology i (for fossil fuel and
non-renewable technologies), and so on. Both loops are balancing as were loops 7a and 7b. In
the simulation model, the TGCs and TECs prices do not directly causally influence the expected
future TGCs and CO, TECs prices. The latter are estimated roughly using the expected demand
indy, the expected total potentially generated wind energy, the expected average TEC price indy,
but with the current specific CO emissions of electricity generated and specific CO emissions of
the different technologies.
Feedback loop 15 — the negative demand elasticity loop ~ : Electricity demand is
somewhat dynamic in this model —although electricity demand structures are not elaborated in
detail- through the negative demand elasticity loop, increasing electricity demand rais
paribus- electricity prices, which in turn, but only after some time and rather inelastically decrease
electricity demand: electricity demand ¥ electricity priceelectricity demand.
The electricity demand is also influenced by some (exogenous) pressures, such as the GDP,
energy intensity and the greenness of the socictal value systen!®.
ceteris
Many other feedback loops, combined feedback loops and technical calculation feedback loops
exist in the model but will not be explored explicitly here, but might be noticed in the simulation
model discussed in the next section.
2 The System Dynamics Simulation Model — In Detail
Here, the structure of the system dynami
of the stock/flow diagram, specific equations and initial values. Although there are many views
and variables in the model, there are also many very similar structures, for example to model
the 9 different generation technologies included in the model. Not all structures, variables and
equations will be discussed in detail in order to avoid repetition, but a revised ver
will soon be made available on the internet.
imulation model will be discussed by means of views
ion of the model
2.0.1 The Wind Power Capacity View
At the right-hand side bottom of figure [3) three related stock-flow structures are visible which
keep track of the wind power capacity installed —starting from an initially installed capacity at
the beginning of 2006 of 33.566 GW (European Commission 2006), the annual total potentially
generated wind electricity and the capital invested in wind power, all three fuelled by the investment
in wind power capacity, which is possibly constrained by a fourth stock-flow structure representing
"The greenness societal value system + effectiveness DSM and REU~electricity demand. ‘The effectiveness of
DSM and REU works marginally on the expected growth rate of the electricity demand.
<total electricity
generate by wind>
vind ecicty
public support wind /'aesigned and generated
ome tect
<cechicty price Gesied “*—~
fff feo
<average price paneer }
HGCe wnt ne” eae *
i; vlatwact
we 1 wood
cat gee
seatgeaifncon > Wid getration Sor cee cost
ei _s ,
-<capital invested in
t erase Sy
<TEC rcveme clean power <average price
‘<TEC revenme clean eetine yod ea per GWh generated previous. KTGC previous
power per GWh beso J ‘year> year>
seoerced>
<erpected TEC revere <espected average
<othoe variable
clean power per GWA prige KTGC iaSy>
cos
eocrted aSy>
peerage lvesmeat
expected electricity he 2
subsidies wind power
ries inSy>
wind power
capacity factor>
<preeoness soctetal expected wind es zy"
L
wind pover capachy
a subsidies
new 7 eapaciy
subsidies wind power
or fe
et private inv we
nee
experience curve
~<elecuicky demand
‘nino must ran
generation>
end year investment
total aew wind
governmental 4 power capacity
subsidies wind power coments
cow ofc
wand 23
ext year” fost ofaew wied
‘apacity previows year
power capacity
So _
intel marginal cost of
wale systen>
pore wee “J total private in cost
a a y ae 7 Wid power parameter vind power
smasinaa L power 4. - ‘
tapected pubic =P ROL new : \ ‘vind power capacity
Support wind wind capacity poreadiockn cmlative 200g
pa siting end other progess rato of wind store wind on
. invesment jower capacity _ capacity commualve experience
siting and other oo 3 reviows year wind power
det and regulatory ee seaman .
smatket and reguetory k
uncertainty and risk = Potential lookup factor *s initial new wind power
amp wind power culate historic g capacity under
= a bo capacity + ~~ construction.
attactivencss to invest
summation oftotal «BREW Wind epacty Pi profiabjity current
anracveness to invest in new 2ioo> { ‘.
oe i wie _e al
gencration capacity setaive aractivensss tof percentage Sa ae
to invest in new wind, contpnt capacity new wind power “1 eapaci installed] wind power capacity | “capaci
peony onreiaite atractvene:s capacity under being decommissioned
\ to invest in new wind ‘construction initial wind power total electricity
capacity istaed generated by vind>
\
generation hours « wind capacky | Heensing and constuction
peryea> Anew wind capacity . need iy bom time new wind power financial loss non
dested % sigament> capacity _-— © generation wind power
Pail Ja + \ ‘total potentially
expected total new wind Oe generated wind %
sew generation power capacity total ne wind power “new potentially lecticky | oki potenbally
required inSy> __eowatissioned capaciy commissioned —se generated wind Bp. fevered wind
Binkinker alechicky lect
capacity factor> / SESS pentane ee initial wind power <gap wind power
‘wind capacicy Gary tin’ plerieb
‘masimam percentage growth se
wind power constnuction [ a fa
industry Capacity of the wind wed power” g Op Wind power potential on
power construction cack ane"? Wed poser eapaciy factor
cromice capes Loch reduction eapacy Took
‘wind power industry ‘wind power industry
frvestment wind LE ™EAPOWET eno wind
es investments:
net private inv cost___
power
new wind capacity>
Figure 3: The stock-flow diagram of the wind power module of the EU-25 wind power model
the capacity of the wind power construction industry. This capacity of the wind power construction
industry is a key driver of the models chapters, also in case of decreasing
electricity demand, which could be seen as push instead of pull. This is not the case here. Here
the capacity of the wind power construction industry is merely damping the growth if the industry
expands too rapidly.
The decommissioned wind power capacity is also kept track of in order to calculate the cumu-
lative historic wind capacity" and from that, the cost of new wind power capacity. This cost of
new wind power capacity is part of a stock-flow structure which introduces gadual™ endogenous
technological change and learning by means of experience curves of the form:
where C; stands for the cost of a wind turbine of one GW sold at moment t and X; for the
cumulative amount of wind power capacity (in GW) ever sold, and e for the experience curve
parameter which is equal to logs (progress ratio). A progress ratio of for example 0.9 means
that the price of one GW sold ~ of initially 900 million €/GW22! is reduced to 90% of its previous
level after a doubling of cumulative sales (International Energy Agency 2000). So the learning rate
equals 1 — (progress ratio). And cumulative experience wind power —which equals the cost of
new wind power capacity over the initial cost of new wind power capacity- could be seen as a proxy
of cumulative learning and experience and is used here to render the technological improvement
of the wind power capacity factor endogenous.
Most of the other variables « in this view are used to calculate costs used to assign
the power generation (wind el assigned and generated and total electricity generated by
wind) and the expected profitability used to assign new pow y investments (new wind
capacity desired) according to the relative attractiveness ive power technologies. The
relative attractiveness to invest in new wind capacity is also influenced by the gap between the
wind power capacity installed and the maximum wind power potential lookup: the closer the wind
power capacity ins ial, the higher the siting and other
investment costs, and the lower the attractiveness to invest in new wind capacity.
The greenness of the societal value systentl also influences the expected public support wind
generation and therefore also the attrac to invest in new wind capacity and the expected
wind power price inSy.
$
tric
alled gets to the maximum wind power pot
2.0.2 Other Power Technologies
The other power technology structures are slightly less detailed and differ somewhat from the
wind energy structures just explained. Technologies considered here are conventional gas-based
power generation (see figure(d), coal-based power generation (sce figure{5) and nuclear-based power
generation (see figure (6), potentially future clean coal power gencration (see figure [7), renewable
biomass power gencration (sce figure |8), photovoltaic power gencration (PV) (see figure [9), and
to a lesser and less detailed extent hydro power generation (see figure [IT) and geothermal power
generation. Other gencration technologies could of course be added modularly to this model if
are considered potentially /sufficiently important for the dynamics of wind energy in the EU-
25 context. Although the structures shaping the development and use of these other technologies
nilar, they also differ from each other when it comes to important details. This is why
ingle subscripted
are very s
these technologi
are all modelled explicitly in separate views instead of in one
10The cumulative historic wind capacity is the sum of the new wind power capacity under construction, the wind
power capacity installed and the previously decommissioned wind power capacity.
'$tructural change could be modelled simplistically by means of temporarily different learning curve parameters.
Recent capital cost estimate of wind electricity generation technologies are €900-1,100/GW for onshore wind
turbines and €1,500-1,600/GW for offshore wind turbines p349).
131f the greenness of the societal value sys tricity demand is reduced 2% per year from
what it would have been otherwise, if the gr etal value system = 50%, then electricity demand is
reduced 1% per year from what it would have been otherwise, and so on.
10
structure. Only the structures and equations of conventional gas-based power generation and the
most notable differences in structure of the other technologies will be discussed.
Another potentially important development in electricity generation —although not an electric-
ity generation technology as such, but which could be seen as one- is the development of storage
technologies (for example hydrogen generation) which could disconnect the variability (intermit-
tence) of demand and supply and allow intermittent technologies such as wind power to grow to
higher degrees.
Conventional gas-based power generation: Four stocks are used in the Thermal Power
Capacity (Gas) view to keep track of gas power capacity installed, the decommissioned gas power
capacity and the capital invested in gas power, and to calculate the cost of new gas capacity next
year of conventional gas-based power generation (see figure). The capacity of the construction
industry is not kept track of because it is assumed that the gas-based power construction industry
is sufficiently big to construct any amount of new gas-based power plants. And the total potentially
generated electricity is not kept track of by means of a stock-flow structure because the technology
is assumed to be mature and not to improve spectacularly which means that —in terms of the
model- the capacity factor does not improve.
The variable cumulative historic gas power capacity is the sum of the new gas power capaci:
under construction, the gas power capacity installed“, the decommissioning gas power capacity
and the decommissioned gas power capacity and is used to calculate the cost of new gas power
capacity ~starting in the base case from an initial averagd!® cost of 500 M€/GW- by means of the
experience curve formula previously discussed and a progress ratio of gas power capacity of 0.85 in
the base case. The expected average generation cost new gas power in 5y is then calculated as cost
of new gas power capacity/(generation hours per year * capacity factor gas * lifetime gas capacity)
+ average fuel cost gas + other variable costs + exp average price kTGC in5y * percentage TGC
required inSy + exp TEC cost gas power per GWh generated in5y.
The average fuel cost gas is estimated roughly from the 2003 OECD Europé!|natural gas price
for electricity generation of 179.3 current US$/toe —fluctuating between 140 and 179.3 current
US$/toe since 1978 (International Energy Agency 2005a\ p79)— and the conversion rate of 1 toe =
1/86 GWh and a US$/€ rate of 1, which gives 2003 OECD Europe natural gas prices for electricity
generation of about 15420 €/GWh (or between 12040 and 15420 €/GWh since 1978).
The proxy variable expected ROE new gas capacity is then calculated as (expected market
price electricity indy * flexibility premium - expected average generation cost new gas power in
5y) * capacity factor gas * generation hours per year / cost of new gas power capacity. The
attractiveness to invest in new gas capacity is calculated as MAX(expected ROI new gas capacity *
(1 + profitability of current gas generation) * attractiveness flevible and storable backup generation
and storable capacity * fuel dependence electricity generation by gas lookup ,0). The MAX(...,0)
function is used in order to make sure that the sum of the attractiveness to invest variables of
all generation technologies is not distorted by technologies with negative ROIs: by using the
MAX(...,0) function, these technologies with negative ROIs are not considered for additional
capacity investments. It should also be clear that the profitability of current gas generation plays
a rather modest role in the calculation of the attractiveness to invest in new gas capacity because
the model does not contain a detailed financial module.
The flexibility premium has been added (only) to the conventional gas-based power structure
because the generation assignment and investment assignment structures do not take the flexibility
of generation or the peak/base-distinction of generation into account, which is a rather important
*
MInitially estimated roughly as 163GW from (European Commission 2006) and (International Energy Agency]
gas turbines (€700-800/GW)
16The OECD Europe and the EU-25 differ in that OECD Europe additionally comprises Iceland, Norway, Switzer-
land and Turkey, and the EU-25 additionally comprises Cyprus, Estonia, Latvia, Lithuania, Malta and Sloveni
17ROI stands for Return on Investment.
ll
s ay font expected TEC cost gas
average fuel cost gas as average fel power per GWh gencreted <percent
percentage of inital fuel cost gas ‘nby> required inS
costa Sk
"Ei ci average fool bout
aie es
generation hours \
inc pee Cn aes
expected average
<canectt) "pe generation cost new gas
cost of new gas
power capacity capacity previous year
<attactive ead / f eee of
ve 8
neabic bain sepecebce sad expected ROI new ccumnalath gas capacity 2006
as eapacity historic gas
capacty
F experience curve i
<firaction of gas based fel dependence of esae ec power BFEWONS cumulative learning =
generation of total te gas based clectkily “P> pt ctivenes. to invest year experience generation by
generation> I i =
generation loclaap in new gas capacity
‘progress ratio of gas.
foeccpeay cumlative historic gas
<profitabilty of currcat relative attractiveness to power vapety
gas generation> invest in new gas capacity
eo
<summation of total
attractiveness to invest in new al
generation capacity> S Wr) _saspower | decommissioning ges
<<eepacity factor new gas power capacity | Capacity installed) power capacity
under construction
gas> “menew gas capacity
<expected total acw-—" gy “sted
generation required
esp gas generation
decommissioned inSy
excess demand of
gas generation eee
‘gas power assiene
and generated
<generation hours
‘per year>
<zeneratioa
hours per
Zaverage price
city factor KIGC= <total electricity
/ percentage generated by gas>
ssignn Sree profitability of current
‘e pepe
tet boteniay — Matghal TGC cost *~
— = generated gas power £6 Seneration <electicity ctechicty
3 capital invested in Se \ price Gesiced
as power | _wrteofls gas Teal generation cost —— <capitalinvestedin fom gas
‘vestment gas investments gas clecticty “FL eas powe
power
“average fuel cost
ed cost o>
ceost of new gas a power Met vatiabie//” cetecscty demand
power capacity> lifetime gas per GWh generated> O85? ‘minus must run
capacity { a oe
« -
aption generation costaa-~Zsunmation of one over
ar katy generatign costs>
‘nancial loss non & Pag
ttl ei pealnie relative attractiveness
generated Py E> power capacity marginal apriori kTGC —_of gas generation
cost gas generation
<average price kIGC —<percentage TGCs
previo
yea required>
Figure 4: The stock-flow diagram of the conventional gas based power module of the EU-25 wind
power model
12
aspect of gas-based electricity generation. This simplistic flexibility premium is a proxy of the
additional benefits reaped by flexible generation technologies and is also seen in reality. The
model is rather sensitive to different flexibility premia as will be shown in the sensitivity analyses.
The flexibility premium used in the base simulations equals 10%.
Conventional coal-based power generation: The conventional coal based power generation
module (see figure|5) differs only slightly from the gas based power gencration module. The initial
values and constants ~such as the initial marginal cost of coal capacity 2006 of 1100 M€/GW28,
the initial average fuel cost coal™, the initial coal power capacity installed of 244GW2 and the
coal power capacity decommissioned, the capacity factor coal and the initial specific CO2 emissions
coal~ first of all differ from those of gas based power generation. A second difference is that the
activeness of new coal generation capacity is not penalised by a high actual dependence on coal
as is new gas generation capacity by a high degree of gas based electricity generation by means
of the fraction of gas based generation of total generation and the fuel dependence of gas based
electricity generation lookup.
A third difference is that if at least 20% of coal capacity i
tive years, then part of this coal capacity idle for 2 years —initially 10%~ is conversed to biomass
capacity by the flow variable premature conversion coal to biomass capacity?! The correspond-
ing capital invested in coal power is written off by means of the variable additional prematurely
decommissioned coal capacity write-offs.
s not used for more than 2 consecu-
Conventional nuclear-based power generation: The total potentially generated nuclear
power by the nuclear power capacity installed -of initially 133 GW (European Commission 2006)-
is lower than the amount of electricity desired from nuclear because of the relatively low variable
generation costs and consequent bid pri ch that the entire total potential nuclear generation
is sold and nuclear could be considered ’must-run’ generation. The generation costs of existing
pacity are relatively low —in spite of additional future nuclear waste storage costs~ because of
the low average nuclear fuel cost and the low average capital cost nuclear generation due to the
largely written-off capital (for about half in the model). The generation cost of nuclear power
remains relatively low as long as the exogenous” variable public support nuclear power seriously
limits the relative attractiveness to invest in new nuclear capacity and consequently the new nu-
clear capacity desired and the new nuclear power capacity commissioned. If thi :
and sufficient new nuclear power capacity is commissioned at recent capital cost estimates of about
2000M€/GW (1700-2150M€/GW (International Energy Agency 2003, p349)), then the average
capital costs of nuclear generation will increase and make nuclear generation somewhat less in-
teresting to invest in —in spite of the experience curve effects and the progress ratio of 0.85. But
even in that case will nuclear generation be treated in the current model as mus
given its rather inflexible
tion is also assumed to be i
ca
is not the
-run generation
character and low variable generation cost. Nuc
sufficiently flexible to be considered an interes
for intermittent generation, hence, it is not causally influenced by the attractiveness flexible and
ar electricity genera-
ing backup technology
storable backup generation and storable capacity. And in the model, nuclear generation does not
18Recent capital cost estimates of several coal-based electricity generation technologies provided by
p349) are: 800-1300 M€/GW for conventional coal, 1100-1300 M€/GW advanced coal, and
1300-1600 M€/GW for coal gasification (IGCC). The initial marginal cost of coal capacity 2006 used in the model
is 1100 ME/GW.
19The initial average fuel cost coal is
stimated roughly from the 2003 OECD Europe steam coal price for
electricity generation of 77.5 current US$/toe (between 62US$/toe and 107US$/toe since 1978)
p78), a conversion rate of 1 toe = 1/86 GWh, and a US$/€ rate of 1, which gives 2003
OECD Europe steam coal prices for electricity generation of 6665 €/GWh (or between 5332 and 9202 €/GWh
since 1978).
20Roughly estimated from (European Commission 2006) and (International Energy Agency 2005a).
2\premature conversion coal to biomass capacity = IF THEN ELSE(coal capacity idle for 2 years/coal power
capacity installed>0.2, coal capacity idle for 2 years * percentage premature conversion idle coal to biomass capacity,
0)
?2'This variable is kept exogenous —although it could be turned into an endogenous variable— because of the major
political and public support required to revive nuclear power in Europe.
13
average fuel cost coal as ital average fuel
percentage of intial fuel cost confoodl
lookup
<Time> <lifetime coal «average fuel © <other variable <percentage TGC
required in5:
cepechys cost coal costs?
mot ‘i a <expected TEC cost coal
<capacity factor power per GWh generated
———— expected generation cost inSy>
coal>
<generation hours pew coal electricity inSy #————-
per year>
<expected averat
price kTGC int
cost of new
<expected electricity".
expected Sectrty oS tap RO! ew coal
. pReckisy: capacity = coal capacity
<profitabillty current a cost of new coal nest year | cost of new coal
coal generation>
q ~_ power capacity Sagacity previous year
Ede nerd Be aod attractiveness to invest i a
activeness fexible and 5 crea
stocable backup generation and innew coal cepactty experience cunniative\_ iia marginal cost of
storable capacity> ave storie coal \coal capacity 2006
a capacity .
__ouamaion oft ave atratvenesto eee revious year uma ering an
attractiveness to investi, OST pew coal eapacty experience generation by
new generation capacity> cumulative histeric co coal
; power capacity
<gencration hours progress ratio of coal ee oo
; Deryear> ~Peneweodcapacky power capacity capacity decommissioning coal
<expected total mye - os gon powercapaciy aes
generation require : | :
inty> ‘ew cod power] _ coal power capacity
capacity under | capacity installed decommissioned
premature conversion
<capacity factor :
et sew coal capacity “Coastction
PTE cad | cs coal to biomass capacity
is Sg. percentage premature
conversion idle coal to
<new coal capacity g
desired inSy fom licencing and lifetime cof eee fraction premature biomass capacity
reassignment> coastruction time new capacity ene ton | toHversion of coal capacity
coal power capacity decommissi steed fraction unused coal
inSy ity z
sperm PT 7 Smcateagcty ie _ poner gern
additional prematurely vr pews yea
decommissioned coal Brecon umsed coal cy demand
capacity writeoffs power generation pines
excess demand coal peneration’
. aoe |
Z coal power assigne
capital invested in
& a ahawes 2 total potentially <ayerage priec\ — electricity
investment coal writeoffs coal generated coal power kTGC> qa
: investment gguerated by cgal> ered!
power invest Ss oo Pe ‘Frc val
4 real generation cost profitability carrent
coal generation
lifetime coal
capacity>
<cost of new
power capacity>
Baa an heen
<eapital invested in
coal power> pace
<summation of one over
generation cos
GW
generated>
<other variable <average fael co
costs coal ctietime coal
<percentage TGf's capacti>
required> a area
atuactiveness
col generation
pos mgt piel
Sa ee vious year>
aptiori generation cost
coal electricity
financial loss non
total electricity
sonrated by cosis 7 eneration coal
- a power capa‘
Figure 5: The stock-flow diagram of the conventional coal based power module which is very
similar to the conventional gas based power module
4
Table 1: Minimal net nuclear electricity capacit
nuclear capacity of the EU-25 in the model
[ I 2006 [ 2010 [ 2015 [ 2020 [ 2025, [ 2030 [ 2050 [ 2100 ]
y in OECD Europe compared to the minimal net
Minimal net nuclear capacity in OECD (133,0) | 112,37 ] 100, 23,67 | 16,87
Europe (Nuclear Energy Agency 2005)
Nuclear power capacity installed if gradually || 133,0 | 120,3 | 106.2 | 93,7 | 82,7 | 72,99 | 44,3 | 12,3
decommissioned by a normal lifetime (40y)
Nuclear power capacity installed f gradually
decommissioned by a longer lifetime (50y)
er capacity installed if gradually 133,0 116,4 98,5 83,9 71,7 61,6 33,0 6,2
decommissioned by a shorter lifetime (30y)
122.8 | 11,1 | 101.2 | 92,7 | 89.9 | 63.3 | 24,7
7 Data for Sweden unavailable from 2010 on, for Spain from 2015 on, for France from 2020 on, for the UK from
2025 on.
yield TGCs (which means that the applicable percentage TGCs required needs to be bought) nor
TECs but does not incur any CO2 TEC costs since it does not emit CO».
The nuclear power capacity i decommissioned gradually by 1/lifetime nuclear capacity
of 40 years in the base © of more rapid —but still gradual- decommissioning,
and of 50 years in case of less rapid decommissioning (see table{I). The table lookup variable ad-
ditional decommissioning nuclear capacity allows the abrupt decommissioning of nuclear capacity
imulate the effect of abrupt phase-outs —as planned by several EU25 countri
of 30 years in c
to
Potential clean coal power generation: Clean coal power —coal-based electricity generation
with CO capturing and storage~ which might become a reality in the coming years or decades,
differs slightly from the conventional coal power discussed before, starting with the initial values
ranging from the amount of clean coal power capacity installed to the cost new clean coal power
n coal power differ from those of
yield revenues or ~as in this model
for conventional coal generation.
The fuel costs of this clean coal are somewhat higher than those of conventional coal: the
additional fue an coal is initially taken to be 10% of the average fuel
the costs of al power could be expected to be even higher due to the ne
pturing and storing. Initially, these cos!
the sensitivity analyses
Another difference with conventional fo is that exogenous structures have
been added such as the initial investment in clean coal capacity to bring about the take-off of this
new technology which is of course not necessary in case of the mature conventional coal capacity.
Although clean gas such as the planned BP hydrogen power plant in Scotland— will most
certainly sce the light too, it has not been included in the model because of the very likely risks
of overdependence on gas and depletion of gas resources which are less problematic in the case
of coal. If clean gas power generation becomes technically feasible and economically viable, then
most likely will clean coal power generation -which has an even bigger potential of clean electricity
generation- too. Today, BP and the Edison Mission Group plan to bring online a low-carbon power
generation plant in California using petroleum coke by 2011. This would open the path to clean
capacity. The structures of the gencration costs of new c
conventional coal power in case of TEC or tax schemes
negative costs for clean coal generation whereas they yield ¢
whic
cost coal. However,
ary CO,
aint, but they are looked at in
are not taken into a
1 fuel technologi
coal power gencration on a commercial scale.
Biomass-based power generation: The biomass-based power view is very
vious clean coal power view. There are two important differen
similar to the pre-
The first major difference is that a static maximum biomass power capacity potential —of
500GW in the base runs~ influenc
amount gencrated) via the gap biomass potential, as well as the investment in new biomass capacity
both the generation cost of biomass electricity (and hence the
23-The initial value of clean coal power capacity installed is ~although no industrial
online~ taken to be 1GW, just to make sure that denominators are not equal to 0.
cured by means of the MAX and IF-THEN-ELSE functions,
lean coal plants are currently
also why several equations
are
15
«lifetime mclear
capacity> fature nuclear waste
storage costs expected generation
cost muclear power
. average nuclear
<expected electricity <capacity factor-ge=- "pected generation ie +— fuel cost
price inSy> ‘mclear> nen ancieat mer vatiable
profitability current eipncceeiaitions costs>
nuclear generation <percentage TGC:
required>
public support exp ROI new <average price kTGC
nuclear power nuclear capacity previous year>
«A
<summation of total
attractiveness to
invest in new
generation capacity>"~a.
attractiveness to invest in
: c cost of new nuclear
new nuclear capacity power capacity ne
year
relative attractiveness Power capacity
<capacity factor iene ik bow!
‘miclear> nuclear capacity
<generation hour 1 1
per year> new muclear '
<expected total new _-- capacity desired _*XPetience curve
capacity previous
initial marginal cost of
muclear power capacity
2006
ear historic nuclear
generation required ¥ nes capacity
Sy> cumulative learning and
ind previous year : e
new miclear power 5 pens seer by
a bears a scopeciy commise progress ratio of clear
io
4 uclear power capacity cumulative historic
we, = power eqpaciy
capacity nuclear power a sz—y|_tuclear power
construction industry power Se pacity
capacity installed Gecommissioning | ee cjoned
nuclear capacity ne
<Time>
ffetime nuclear ae ,
F z decommissioning average capital cost
<new nuclear capacity capiicky r :
desired inSy from 7 imucer copay sg
reassignment>
<generation hours
per year>
<cost of new nuclear we Se <lifetime nuclear
writeolis age apacity>
Sy investments
power capacity>
additional fielcost clean coal
additional carbon
capturing and storing—w. 2Verage fuel cost
costs
___— =
“peneiation hours clean coal elect inSy
<average fuel cast
16
other variable
costs>
clean power per GWh
generated inSy>
7 clean coal coa>
Nr a 1 eo eC revenue
exp generation cost new —_
expected average
~~ *
peiseingy> exp ROT new clean Pre
coal capacity = acky next 3
<profiabilty curent cost of new clean coal cast of new clean coal
clean coal generation“ sc activeness invest in power capacity
satractiveness
ible and ye new clean coal capacity
experience curve
<summation of total
power
attractive
relative attractiveness to
invest in new cleaa coal
aciey
5s to invest
‘previous yea
»
08 <generation hours
<expected ictal new PET Yea,
parameter clean coalf cumulative historic
clean coal capacity
progress ratio of clean “comulotive historic clean
capacity previous year
initil marginal co
Jean coal capa
cunmiative learning and
experience generation by
clean coal
initial decommissioned
clean coal power
oan reared coalpower capacity coal power cepacity eae
Se eS new clean con apacily
capacity desired
a decommissioned
new clean coal ¢~) oy" y, | clean coal power| clean coal power|
<aew dean coal capacity capacity pane orto ae ae
desired indy from commissioned new clean coal aoa capaci
cassia - _
poner capacity pe ey
oe constuctos \ *
lifetime clean coal
clean coal generation capacity
initial investment in ba 2 ss <generation
Icencing and construction issioned inSy 8
ean cal capecay iegles emecdvoegy eg ery = hows pe
capacky at>
dae 9 cfeleo gen by excess demand clean sein decimal
inpvestment ee oa coal generation ‘minus must ran
leer coal generation hours 7:
Bower per ve clean coal power
en assigned and a Ph
capital invested in total potentially average price otal =
cleen coal power * genertedlemcod KIGC> __eeterated by deen
es vite clean” SS Oe Sf desired
coal investments
<cost of new clean coal
power capacity>
«lifetime clean coal
capacity
pane TEC cost clean
coal power per
GWh generated>
<TEC revemue clean
power per GWh
generated previous
TA
financial loss
aoa-generation clean
coal power capacity
seal generatioa cost
clegn coal electricity
a
ts ‘gencration a
clean coal electricity
from clean coal
profitability current
clean coal generation
capital invested in
cleaa coal power>
<electricity
‘price>
<suuunation of one over
generation costs>
<lfetime clean
capaciy>
cage fuel
cost clean coal>
a
relatract clean
coal gen
<opgarte
‘previous year>
Figure 7: The clean coal view of the stock/flow diagram
17
via the expected gap biomass potential, by increasing the (expected) average fuel costs of biomass
by means of the average fuel cost biomass via gap lookup. The average fuel cost biomass via gap
lookup assumes in the base runs: ((-1,5), (0, 1.5), (0.5, 1), (1, 0.8) whi
that the average biomass fuel cost is 80% of the full price when there is no biomass-based electric
generation capacity in place, is 100% of the full cost when half the maximum potential biom:
capacity is in place, 150% if the maximum potential capacity has been reached, and even more if
it peaks above this initial maximum potential capacity. Thi
of other structures curtailing the biomass capacity expansion~ to severely oscillating inv
behaviour -not only for investments in biomass capacity but for other generation capacities too-
when the biomass capacity —initially 11.549 GW (European Commission 2006)- approaches and
exceeds the maximum potential, imposed by the maximum available biomass fuel. But a direct link
has been added to the model between the expected gap biomass potential and the attractiveness to
invest in new biomass capacity in order to damp these extreme oscillatory patterns. Adding this
structure leads to a smoother evolution without heavy booms and busts. It could be argued that
real-world information about the future availability of biomass will become an important aspect of
the decision to invest in biomass power capacity once the maximum potential is being approached.
The second major difference is that 10% of the coal power plants which are not used for more
than two years and which are therefore decommissioned before the end of their lifetimes, are trans-
formed to biomass power plants —by the variable premature conversion coal to biomass capacity—
at the cost of the additional prematurely decommissioned coal capacity writeoffs, which is lower
than the costs of newly build biomass power plants of about 2000M€/GW (1,500-2,500M€/GW
according to the [International Energy Agency (2003, p349)). These conversed plants are assumed
to be 100% biomass plants —which is of course a simplification of reality and therefore receive the
full amount of TGCs and TECs.
means
structure alone leads —in the absence
ment
Solar Photo Voltaic: European Photo Voltaic (PV) power has been growing by about 35% over
the last couple of years in spite of the relatively high marginal investment cost of about 6,000-
7,000 M€/GW for distributed photovoltaic and 4,000-5,000 M€/GW for centralised photovoltaic
(International Energy Agency 2003] p349). But these costs are falling rapidly. Thin film PV
power technology might —within a couple of decades~ become a major power generating technology.
However, the future large scale deployment of this technology is not certain. As for now, costs are
still prohibitive for a large scale deployment. This makes that PV power ~contrary to wind power~
is currently only viable in niche markets or in case of extensive governmental subsidies. Given
the high uncertainty/riskiness, a scenario/controlled simulation approach is opted for here. The
development of solar PV power generation and its impact on the development of other generation
technologies is included endogenously in the model, but it is kept exogenous to the rest of the
model which means that the developments of the other gencration technologies do not causally
impact the development of solar PV power generation. This comes down to the assumption that
solar PV power will be mostly installed in a decentralised/distributed mode, independent from
centralised generation. The extension of the model with real competition for capacity extensions
with the other gencration technologies would be very interesting, especially in case of (expected)
increasing fuel prices.
Again, the same type of variables —but slightly different conne:
First of all are the stock variables PV power capacity installed (starting from an initial PV power
capacity installed of 1.010 GW ( es Coe) and ~after an average lifetime of
PV panels of 20 years in the base case~ decommissioned PV power capacity used to monitor the
capacity and to calculate the cost of new PV power capacity decreasing following the experience
curve with a progress ratio of PV power capacity of 0.75 in the base case. This cost of new PV
power capacity is used to calculate the capital invested in PV power which is monitored separately
because of strong cost-reducing experience-curve effects, and which is used in the calculation of
the average generation cost of PV electri
ctions and functions~ are used.
?4The average annual percent change of solar PV between 1990-2003 was 34,5% in the OECD EUROPE and
35,9% in the EU-15 (International Energy Agency 2005b).
expected TEC cost coal
<expected average tia
18
average fuel cost biomass as
ie ‘centage of initial fuel cost
2 we power per GWh generated “other variable Per <Time>
price kTGC in5y> insy> costs i
Afetine biomass ee average fel cost
capacity expected average faclet——____ biomass via gap lookup
a es AiG cost biomass indy ~
iomass>
expected electricity
price inSy> sp RD ines
sy biomass capacity ;
profitability curent ay
biomass generation
catracveness Rexble and Stactveness to invest in
storable backup generation and” 7” ew biomass capacity
storable capacity expetience curve
cost of new biomass
cost of new biomass
capacity previous year
initial marginal cost of
biomas
\
power capacity
ity 2008
i expected
ameter biomass | cumulative historic pected 6p
wumalion cftotal_ _selaive etiacivensss PME mass | comubavs sion : biomass potential
atractiveness to invest in to investia new nas anda leaning ee
new generation capaciy> ‘biomass capacity ious experience generation by
<ecnsratioa hours progress ratio of cumulative historic se,
peryear> “te new biomass Soeiass poier tie pone rene
expected total new" a capacky =,
reverand Ie 2 oa
copacity factor new biomass capacity bias pa — ‘biomass power] asian biomass
i new biomass power| cepacity installed | ~aecamumissioring cepecity | power capacity potential
w bi capacity under |“ biomassponer | decommissioned
deednSy fom constuction See ee
‘aia <prematare comveision f fieime biomass
cof new biomass coal to biomass capacity exp biomass generation capacity ie
power capacity licensing and decommissioned inSy “ zc
; coastracson time es
additional prematurely “new biomass :
on £3 decommissioned coal power capacity/ capacity factor arpeeaae Oy ees
‘avestment capacity writeotis> bene biomass generation ee potenti
‘biomass ‘tines power ke, generation»
sever hapd et eee, F,
average price clectikcity desired
total potently —KTGC> —_<total electicky ‘from biomass
‘generated biomass generated by biomass> oy.
cokes power ice <Time>
pre
Ps si <average fuel cod
‘biomass clectricity f
power per GWh
generated>
<TEC revem:
previous year
financial loss non
‘generation bicmass
power capacity
Figure 8: The
view of
ICTEC reverne clean
GWh generaied
-”
2 Softer <summation relative attractiveness
vaviasle of one over 9 biomass generation
costs: genscation
costs?
‘average price KTGC
preyious year
priori generation cost”
biomass electricity
the
tock/flow diagram
19
experience curve a
parameter PV progress ratio of PV
power capacity power capacity
power capacity
v, previous year © '=
cumulative historic PV cost of new PV’
power capacity power capacity.
decommissioned
PV power
be |e gs PV power
capacity initial marginal cost o
new PV Sa | a decommissioning |___capacity DY pov cae
a construction PY capacity = :
mnstruction time pe \
growth x3 I, new PV power new investment ff
new f capacity capac 6 PV power
soon i 4
YN sap PV potential PV snew PV capacity | capital invested
power capacity commissioned> | in PV power
licencing and potential wn PV.
construction N oF pans ¥ |
ie sev" maximum -witeofis PV
potential
power capacity PV average generation power
capacity cost PV electricity investments
- cumulative leaming and
solar PV experience of PV power
new solar PV electricity decommissioning solar sree vsjeamasas
: generation > <TEC revenue clea
electricity generation |_& PV electricity generation power per GWh
<average price generated>
expected potential PV. “initial capacity “average price = ©
<generation ones factor PV KIGC>
hours per year>
capacity
factor PV
Figure 9: The solar PV generation capacity view
The increasing cumulative learning and experience of PV power capacity —cqual to (1 - cost of
new PV power capacity / initial cost of new PV power capacity)— slightly increases the capacity
factor PV as well as the maximum potential PV capacity from the initial maximum potential
PV capacity. The increasing capacity factor PV increases the new potentially generated solar PV
electricity of new PV capacity commissioned and therefore also the solar PV electricity generated
—monitored separately be: city factor, and effectively generated as *mus'
run generation’ in this version of the model5| which decreas
electricity.
The increasing maximum potential PV capacity also increases the gap PV power potential
which leads in turn to more new PV capacity commissioned which drives —after a licensing and
construction time of new PV power capacity of 1 year~ the new PV capacity under constru
the new potentially generated solar PV electricity and the new investment PV power, and so on.
Here again, the new PV capacity commissioned does not compete directly with the centralised
generation technologies for new capacity to be commissioned, so there is no feedback loop with
the larger model at this point. The interaction with the larger model might be improved by
introducing a variable cost of new PV power electricity in5y interacting in a feedback loop with
the expected generation price, partly driving the expansion of PV power.
The average generation cost of PV electricity is also influenced by the average price kTGC and
the TEC revenue of clean power per GWh generated from the rest of the model. Several variables
from the PV view also influence the variables in the rest of the model: solar PV electricity generated
cause of the dynamic capa
s the average generation cost of PV
25PV power does not compete for supplying part of the electricity demand: it simply delivers all electricity
generated as distributed must-run generation. This means that there is no feedback loop to the other generation
technologies at this point.
20
Graph for PV power capacity installed Graph for cost new PV power capacity
1,000 6B
500 3B
0 i 0
2006 2030 2053 2077 2100 (2006 2030 2053 (2077 2100
Time (Year) Time (Year)
Graph for solar PV electricity generation Graph for capital invested in PV power
2M 600 B
1M | 300 B
0 0
2006 72030 2053 2077 2100 2006 2030 2083 2077 2100
Time (Year) Time (Year)
Figure 10: The time-evolutionary behaviour of the solar PV structure
influences the must run generation and the total amount of electricity (potentially) generated; PV
power capacity installed influences the percentage intermittent 4
expected PV generation in5y influences the exp TGC supply inSy. Initially it is also «
that the PV power gencration does not influence the el
grid because of its (
sumed here
'y market price of the centralised
amed) purely decentralised /distributed character.
Hydro Power Capacity: On-shore hydro power is modelled rather simplistically in this model:
large, small, micro and pumping capacity are treated without distinction and the further de-
velopment of hydro power is independent from the development of other generation technologies
(see figure [IT). The latter simplification is made because
dro is about reached which means that further growth will most probably come from dis|
the maximum capacity of large hy-
tributed
(small, micro and some pumping) hydro capacity which will mostly be added by small, local
players in a decentralised mode. The hydro power capacity installed grows with the annual new
hydro power under construction which equals the assumed growth rate new hydro of 10% times
the hydro power capacity installed times the gap hydro power potential of (1-hydro power capacity
installed / maximum potential hydro capacity). The assumed maximum potential hydro capacity is
-in the base case~ taken to be double the current hydro power capacity installed of 131.440 GW
(European Commission 2006). Recent capital cost estimates of hydro electricity generation tech-
nology of 1900-2600M€/GW (International Energy Agency 2003) p349) and an assumed lifetime
of 40 years are used for generation cost calculation purposes. Future hydro electricity generated
equals the hydro power capacity installed times the current capacity factor hydro and the generation
hours per year.
Although hydro power is almost exogenous to the rest of the model, it still has an influence on
the rest of the model, more precisely on the must run generation and thus the electricity demand
minus must run generation and the electricity desired from other generation technologies, the
TGC supply, the expected TGC supply inSy via the expected hydro generation inSy and the total
expected potential generation indy.
26 sensible extension of the model might be to split them in order to distinguish pumping capacity and part of
large hydro power —which could be used to absorb intermittent/variable output generation— from all other hydro
power ~some of which is variable (seasonal) output generation. They are also often treated differently in terms of
TGCs received.
21
loge cost of new hyo Graph for hydro power capacity installed
+ power capacity 399 ——————
+ hydro Ng pe
Be Sue ' aa
licencing and generation” ~®real generation
construction = cost hydro 1S
time new hydro > electricity /
power capacity =
hydro power = 100
capacity one 30: 2053 7077 3
new hydro power |__installed _| decommissioning amg A030 wie sayy 20 7100
under construction + hydro ¢
i : - io -
growth vate | 2p hydro power Graph for hydro electricity generation
Few bye potential initial hydro power —_ifetime 600,000
+ capacity installed hydro
power
maximum potential ——
fivdro cqpaciy 4 40500: |— ==
oF capital invested
investment hydro| inbydro power} —_writeofs 200.000
power hydro power 2006 2030 2083 2077 2100
<cost of new hydro Time (Year)
power capacity>
Figure 11: Simplified stock-flow structure and resulting dynami
capacity installed
of the European hydro pow
Other Generation Technologies: One other relatively less important generation technology is
included here, namely current geothermal capacity of 0.695 GW with
a geothermal power capacity factor of 91%, a recent capital cost estimate of geothermal electricity
generation technology of 1,800-2,600 M€/GW (International Energy Agency 2003) p349) and an
ars. This structure could possibly be further developed.
assumed lifetime of 20 5
Hydrogen Technology: Hydrogen technology is not so much an energy generation technology,
but it could seriously impact the electricity generation sector in at least two ways.
A real breakthrough of hydrogen would first of all make available the necessary technologies
to disconnect the double variability/intermittence of demand and supply —which would allow
variable-output technology to penetrate with much higher percentages. In the model (see figure
(12), the exogenous variable year real hydrogen or storage breakthrough allows to switch on the
switch hydrogen or storage breakthrough (from 0 to 1) which turns both the impact of percentage
variable output capacity on relative attractiveness to invest in new wind power capacity and the
attractiveness flexible and storable backup generation and storable capacity back to 1.
The second serious impact of a real breakthrough of hydrogen would be a temporary increase
of the electricity demand growth increasing the electricity demand. This again could be introduced
exogenously in the model by means of the variable additional growth rate electricity demand by
hydrogen breakthrough added to the expected growth rate electricity demand.
2.0.3 Generation Assignment
The generation is assigned in three steps. First, the ‘must-run’ generation is automatically assigned
to generation technologies which are either distributed (for example solar PV ¢ ty generation,
hydro electricity generation and geothermal electricity generation) or are inflexible with very low
sts and have therefore very low bid prices (for example nuclear power assigned and
Then, the electricity demand minus must run generation is assigned to the remaining ‘may-run’
technologies —not included in the ‘must-run’ generation— based on gas, coal, clean coal, biomass,
signment is taken to be inversely proportional to the
and wind power generation. This first real a
22
impact of percentage variable output
capacity on relative attractiveness to
invest in new wind power capacity
oe impact of perc var output cap on rel
psaiaagtnceaeakes attractiveness invest in new wind ceils
storage breakthrough power cap ifno hydrogen coy lented
S capacity jastalled> :
eakthrough ~ -
switch hydrogen or percentage “ .
storage breakthrough Stent Pia
a ae capaci capacity i
<Time> attractiveness flexible and Storable 7 ra
backup generation and storable
capacity if no hydrogen breakthrough
attractiveness flexible and
storable backup generation and
storable capacity
Figure 12: The potential influence of hydrogen technology on intermittent and non-intermittent
power generation technologies
a priori generation costs —the generation costs if fully assigned and with TGC and TEC prices of
the previous year~ of the respective generation technologies. In the case of gas based generation,
the relative attractiveness of gas generation is for example equal to 1/a priori generation cost gas
ration
tricity / summation of one over generation costs. This relative attractiveness of gas ge
times the electricity demand minus must run generation makes up the electricity desired from gas,
which could be smaller, bigger or equal to the total potentially generated gas power. The amount
of electricity desired from gas smaller or equal to the total potentially generated gas power is
assigned in gas power assigned and generated, the over-
of gas generation which is —together with the exce:
reassigned.
All over-allocated amounts are summed in the variable total e1
which is r
assigned rest is held back as excess demand
demands of the other generation technologies~
ess demand to be reassigned
signed by means of the alloc p and market p functions2”| ‘allocation by priority’
to the remaining potential generation after the first (real) assignment —available gas
generation after the first assignment in the case of gas based generation (see figure [13). These
remaining available amounts of generation are offered to the allocation mechanism by means of
the subscripted variable total available generation possibly to be reassigned after first assignment.
The width29l taken here is one over the difference in price of importance of €500/GWh. And the
priority of each generation technology is again equal to one over the a priori generation cost of
that technology. The resulting generation allocated per technology is then unsubscripted in the
total excess demand reassigned to generation alloc —in case of gas based generation in the variable
s demand reassigned to gas generation alloc. And the total electricity generated by cach
may-run’ generation technolog made up by this total excess demand reassigned to
generation alloc together with the previously assigned power assigned and generated.
A generation shortage arises if the total excess demand to be reassigned is greater then the sum
of the amounts of available generation after the first assignment.
This particular structure —which is only one of many possible structures~ influences the be-
haviour of the model. But alternative structures are not explored here.
structu
27*to allocate a scarce supply [electricity demand in our case] to a number or requests [generation supplies in our
case] based on the priority of those requests’ (Ventana Systems 2000)
28See Appendix E of the Vensim Reference Manual:
29The width ‘specifies how big a gap in priority is required to have the allocation go first to higher priority with
only leftovers going to lower’ priority (Ventana Systems 2000)
23
total excess demand total excess demand total excess demand _ total excess demand total excess demand
‘total potentially teassigned to gas reassigned to coal —_reassignedto clean coal teassigned to biomass reassigned to wind
gencrated gas power> generation alloc generation alloc generation alloc generation alloc generation alloc
<gas power assigned
~. _
and generated> Wm avaliable gas generation i
<coal power assigned —_ after the frst assignment "generation allocated total excess demand
mE per technology ee aasienad Renin haede
total potentially a ° cost be dete
generated coal power> PO Svalable coal generation sq. we prin gene
after the fist assignment g ~ total available j cost coal elect
Atal potently ee ~<aprici generation e33
seremtedcean cool, raise cleancoal 7 cevfistenigunc [oa _ semncontlety
ape ed cow generation after the first <apriori generation cc
aigedied eA v . biomass electri
? ae Boies vaenty oti generation co:
na avaleble biomess diference in price of "par technology eae
total potentialy generation aler the importance
generated biomass" assignment
WE vA total excess demand generation
'\____»!
iomass power tob ed shorta
assigned and available wind generation ee a ee w:
generated> after the fst assignment a ~
total potently ZesGct dein CN excess anon coat <ersea actin D ces demand" ebacess imandoé
SS coal generation> generation> biomass generation> nuclear generation gas generation>
wind electricity ‘total potentialy excess demand wind | <cleciricity desi
puree generated wind — sel Wd geht) dese
assigued and geucrated> praree generation fiom wind>
Figure 13: The Stock-flow diagram of the gencration assignment to the may-run generation tech-
nologies
2.0.4 Electricity Demand, Demand Forecast and Supply Forecast in the Model
The stock variable electricity demand —starting from an initial value of 3179000 GWh in 2006
(European Commission 2006)— incre:
expected additional electricity demand which is the product of the electricity demand and the ex-
(and potentially decreases) by means of the flow variable
pected growth rate electricity demand (sce figure{I4). The expected growth rate electricity demand
is equal to the GDP growth EU * electricity intensity EU growth + annual percentage decrease
electricity demand by spontaneous DSM and REU + annual percentage
mand by forced DSM and REU - annual percentage decrease electricity price * pric
additional growth rate electricity demand by hydrogen breakthrough. The bas expected growth
rate fluctuates around the 1,9% electricity demand growth rate observed in OECD Europe and
the EU-15 between 1990 and 2003 (International Energy Agency 2005b) (which is higher than
the 1.3% projected by the|International Energy Agency (2004, p218219)| to 2030 for the EU-15).
The GDP growth EU consists of a GDP growth EU trend factor and a smoothed random normal
factor GDP growth EU randomised which might be used for scenario and sensitivity analy:
The expected electricity demand indyea ity demand times the ex-
pected growth rate el
expected electricity demand indy augmented with a supply margin desired which is assumed to
quals the current electri
lectricity demand®. The total potential generation required inSy is then the
depend on the scenario variable degree of real EU-wide competition determined by the lookup
able degree of real EU-wide competition lookup. This total potential generation required indy
s the total expected potential generation indy gives the expected electricity shortage inSy. The
expected total new generation required indy is then calculated as MAX (ezpected electricity shortage
in5y/5,0)*(1+additional construction margin). So, the difference between expected demand and
expected potential supply within 5 years gives rise to additional investments in new generation
capacity.
The aforementioned total expected potential generation indy is the sum of the potentially gen-
crated electricity in 5 years by all generation technologies, which are approximated rather well
by the respective potential generation plus 5 times the newly commissioned potentially generated
electricity minus 5 times the newly decommissioned electricity generation (sce the right hand side
structure in figure (14). Figure [15] shows the similarity between the approximation 5 years before
mint
24
<new potentially <iotal potentially <old potentially
degree of real EU-wide generated wind generated wind generated wind new gas pow
Sompettina lockup. So clectricity> clecticity slectricy> capacity under
<shortage expected total expected gas power"
per degee of real papaiey a coca
EU-wide es eneatasy sl Be epacity inSy apacity instlled>
Sy expe acy factor <decomanissioning gas
A GDP growth GDP growth EU wind generation inSy > ‘capacity>
<Tine>— ODP. supply mace power capacity
EMtced seolied sesioutes desired ge <generation hours <coal power
7 J electricity Y ‘expected potenti £05-— per your capacity installed>
nena pire gt decrease intensity rota potential, capected gencration
al peelings desteese FU growth geuctatoa requiem Slecticty pected coal power g _<ecomissioring cos
GDP gowh EU ‘iaSy shortage ny ee cepaciyin$y povter copacity>
{ wa ¥ expected potential coal -<new coal power
reenness soci geacration inSy : capacity unde
“eemnesfocetl TR scdgomic___g enpecedebcicty ‘lal npeged scopy fc caret oe
a clectricty demand demand inSy BEHTEHORESY potenti on coa new clean coal power
gual percage ‘coal generation inSy sapacty mick
‘demand by expected clean coal
spontaneous DSM ceaniay power capacity inSyf_—_caparity nstal
adie / ae \ SEE Se jon
elasticity | electricity demand expected potential <ecrcration hours
<anmuel percentage additional growth rate a ae per year “new biomass power
decrease electricity electricity demand by -t—<Timc> expected capacity under
price> ydrogea breakthrough potential PY construction>
generation ind expected biomass ctiomass power
‘expected expected potential ver capacity inSy capacity installed
jessie ‘generation potential hydro \, Muclear generation inSy i -<decommissioning
a aed biomass power
biomass> capacky>
raport Sati pa en eet
mip ee power capacity i pipes lent 4: ter
geuerated> ver capaci inSy“Z uncer consrinn>
<hydro electricity <sola PV electricity scration ‘nuclear power
generation> hows per <capa apacity installed>
generation>
nuclear capacity>
Figure 14: Electricity demand, expected electricity demand indy, total expected potential generation
in5y, and expected total new generation required in5y
25
10M
2011 2033 2056 2078 2100
Time (Year)
Figure 15: Comparison of the total potential generation and the total expected potential generation
indy with a delay of 5 years
and the real total potential generation.
2.0.5 Capacity Investment Assignment
The assignment of investments in new generation capa
assignment discussed previously. First, some technologie
tributed solar PV based and hydro based generation caps
ity are rather similar to the generation
are ‘exogenously’ invested in —the dis-
Then, the expected total new generation required indy is assigned ~in the first real assignment—
proportionally to the relative attractiveness to invest in new sity which is equal to the at-
tractiveness to invest in new capacity of wind-based, gas-based, coal-based, biomass-based, clean
-based, and nuclear-based®! power generation capacity, divided by the: stammation of total
st in new generation capacity. The relative attractiveness to invest in new
power capai s the expected total new generation required indy divided by the product of
the respective capacity factor and the number of generation hours per year gives the new capacity
desired of cach of the generation technologies.
Again, this particular structure —which is only one of many pos
behaviour of the model. But other structures are not explored here
However, these new capacities desired sometimes exceed the capacities of the respective power
plant construction industries in which the capacities actually commissioned are limited to
the capacity of the power construction industry. sometimes the c
power capacity commissioned, The rest is channelled via the variable e:
capacity to the variable total ¢ demand new generation indy (see figure 16). This total
demand new generation inSy is redistributed by means of the alloc p and market p functions
over the generation technologies with remaining spare industrial capacity on the basis of the
respective relative attractiveness to invest in new power capacity of the different t
their available new generation after the first assignment. This results in the assignment of this
s new capacity demand to the new capacity desired indy from reassignment variables which
are additionally commissioned.
ble structures— influences the
for the new wind
cess demand new wind
nologies and
exces
2.0.6 Electricity Price
Electricity price levels and specific price structures differ markedly within and between EU-25
member states due to different technological mixes, market structures and other local circum-
stances. Any pricing mechanism in the model would be a severe simplification of reality because
(i) the model is aggregated on the EU-25 level, (ii) it is only simulated a couple of times per year
and not per quarter of an hour, and (iii) does not deal with different (electricity) markets. And
S0if the
31
is sufficient public support nuclear power
new wind power capacity commissioned = IF THEN ELSE(new wind capacity desired > 0,IF THEN ELSE(new
wind capacity desired < capacity of the wind power construction industry, new wind capacity desired, capacity of
the wind power construction industry), 0)
26
-<genesation hours
ra ° Ne a
sew ens cqgacky 4-1 Cod capaci le ‘Somos Sag sca aocc capa
<eapaciyf
capacity of the ao available new gas desired nS aces desired inSy ftom desired indy from desired inSy from desired inSy from desired inSy from
poet ctte my eee ie = reassignment seassignment a ca reassignment
industry asigument alin fico eat 2 <wind powe capacity fact
new gas capacity mri oat capacity factor
Sesirec> ew gas generation new coal generation new clean coal generation NeW biomass generation nev wind
w coal capacity desired nSyfiom — desiedinSyffom —desiedinsy fom dested Sy om generatondesired inSy
decked resssipmedt reassiganent reassiounent reassignment from reassignment
available now coal
| pe genzration after the fist
assignment
capacity of the coal
ower construction
wz
Tew generaticn in total neve generation
oc techolosy ag ieeel
total oealable new generation \
eteation SY possbiyto be reassimed Wdthespsciy | prorqy new capaci
OTL DE ge after fcst assimment reassionment reassignment
available new clean coal
generation after the frst
capacis of the biomass available new biomass
Saver consiictas, —Pegenertion for the Set ox’, ser generaon
‘gancy sssigauect, sesesigmmectprocty ractiveness to
oy ee attractiveness new Ber techiolo iavest in new biomass
ernerationinSy capaci
caclatve nractveness to
- total excess demand. generation shortage invest in new nuclear
available new wiod ew generation inSy ~~ =e generation ink capacits>
gencraion atc the fest
rad power “generation hours a
capacity factor> eae pari
pau \’ _UITBREIDEN TOT ANDERE demand nes
capacity of the muclear available new nuclear GENERATIETECHNOLOGIEEN ai
power construction —t generation fe the fest
soy asseament
gees ithe Sint sass a
new miclear eo clan col eapaciy> "> atactvenss to invest ia at “atvenes fo vest
capaci decked fe new generation cepacity omass capac
attractiveness to invest <attractiveness fo invest <atractiveness to invest “attractiveness to invest
imnew wind capacity> —innew gas capacity in new coal cepacity> in new nnclear capacity>
Figure 16: The new capacity assignment and reassignment view
27
<financial loss non <financial loss a
st ‘<tinancial loss
annual percentage a ————— electricity price apostetiond generation coal power aon-generation clean 2
decrease electricity electricity — coal power capacity> "om generation
price market price gas power
vious perio capacity>
Preven pied cclamount of Waallossnon :
eneration
y electricity generated> financial loss
financial loss biomass power
* capacity>
markup. total merginal loss win
Lowes = ‘non generation
“action of biomass ,
of total electricity a mass elec
aptior electricity tion te aposterior eleciicty
<aprioni generation cof AGAR. “8 <fiaction of wind of total", Market price
biomass Sc electricity generated> geo :
<apriori genera <fraction of clean coal generation of
cost gas clectricity> total clecticity generation» .
<apriori generation <fiaction of coal based ?
cost coal electricity>, generation of total wind electric
<aptiori generation cost stocky : <real generation
wind clecticity> s electricity>
<fiaction of gasbased —_<faction of hydro of total <fraction of melear of total <real generation cost
generation of total gene; geaeration> clean coal electricity>
electricity price
previous period /
expected electricity
price inSy
<apriori generation cost
clean coal electricity>
electicity generation
Figure 17: Three price structures and the resulting el
the annual percentage decrease electricity price
'y price, expected electricity price and
different pricing structure influence the market. This is why the sensitivity of the model
and strategies to other price structures is explored as well. Three simple pricing mechanisms and
their mixtures
‘ould be tested for (see also figure {I7): (i) an a priori electricity market price
-which could be seen as a proxy for a market price~ calculated as the maximum of the a priori
generation cost of gas, coal, wind, biomass and clean coal power if their generation exceeds 15% of
total electricity generated times a markup and a shortage markup (activated in case of generation
shortages); (ii) an a posteriori electricity market price which could be seen as a proxy for benev-
olent monopolist pricing~ which is the sum of real (ex post) generation costs per technology times
their respective fractions generated of the total generation times the markup and the shortage
markup plus the total marginal loss of non generation; and (iii) an a posteriori electricity market
price of the previous period which is nothing more than the a posteriori electricity market price
lagged by a year. The annual percentage decrease electricity price (times the price elasticity) in
turn influences the expected growth rate electricity demand in the demand view. And the expected
electricity price indy is taken to be equal to FORECAST (electricity price, 5, 10). And the elec-
tricity price structure used here is equal to 1/3 the apriori electricity market price plus 1/3 the
aposteriori electricity market price plus 1/3 the aposteriori electricity market price of the previous
period. And there is a general electricity price markup of maximum 5% in case of shortages.
2.0.7 Emissions and Schemes
The 'Emi contains variables and calculations concerning the Tradable
Green Certificates and Tradable Emission Certificates (see figure [19)
The average TGC price, average price kTGC, average price kTGC previous year and the
expected average price kTGC indy are calculated from the TGC suppl! from green generation
and TGC demand from the total amount of electricity generated times the exogenously enforced
percentage TGCs required, and the exogenous minimum TGC price and maximum TGC price.
Figure [19] shows the part of the model where the TEC revenues and c
the (specific) CO» emissions are calculated from electricity generated, the specific CO emissions
and marginal fuel uses and the TAX/TEC price lookup.
sts, the fuel uses and
32In which clean coal is not included here.
28
expected potential PV <espected geothermal
expected average <expected electricity
generation inSy> power capacity inSy> price kTGC inSy demand inSy>
<espected potent
hydro generation in5y>— a espected TGC. expected average. expected TGC
od potential apiisy ToC ae py demand inSy
sind generation inSy>
“
Serpe ymin expected mitimm expected maximmum percentage TGC
none Geert TGC price in5y TGC price inSy required inSy
minimum TGC maximum TGC percentage TGCs
wore percentag
average price KTGC a ein lockup pricelookup required lookup
RE ENIOC rage pce
, minemmTGC — maimumTGC percentage TGCS
oe Py price ice required
toa poten as supphy— a" age ICC 160 demand
er ecticts> we
decwiciy>
hydro electricity othermal total ec al ammount of
generation> electricity generation» g;
erated by biomass> icity generated>
Figure 18: The emissions and schemes view of the stock/flow diagram
cumulative amount of
CO? avoided by wind
« ca Te expected TEC
power generation | total annual amount of CO2
revenue clean
eciic CO2 emissions of
specific CO2 emissions ™ power per GWh
woided by wind power ~t— mde acral
eed ee of electricity generated teva gence generated inSy
total electicity generation er a ~~ TEC revemue clean power
generated by wind> total amount c total CO? TEC ae ae per GWh generated
electricity generated> is Bower Pe y
initial spectic fue! use Hectncdy ctr red cored previous year
elec generated by coal Leaeadalive wicedng aiid fill spouiie CO \ ae
8 eg SD MRCEseeratonb— emissions coal TEC cost coal power average TEC price
coab>
pec GWh generated previous year
generated by coal Te.
otal electricity CO2 emigsions elec 4 specific CO2
TEC cost clean coal mveser
gencrated by coal> generat emissions cont cer per Gh “TEC price
coalnge dectricty generated »
generation ~xq__ <total electricity co2 pereange s
‘generated by clean elec geherated =~ (0)? emissions TEC cost ns ones TOTES
coals 7 cea coal pai per GWh generated price lookup
additional percentage coal y clean coal a
use clean coal generation a
exp TEC cost coal power
marginal fuel use elec. cas use electricity CO2 emissions elec | specific CO2 an
ae ted inSy
generated by gas.” generation nerated by gas emissions gas____ Pet GWh generated inSy
total electicdy
initial specific fuel use generated by ae
lee generated by cat
exp TEC cost gas power | ~z
« TEC
. ber GWh generated insy “POSSE
cumulative leaning and iii spect COd ial
experience gencration by gas: euuissions ges
Figure 19: The COz emissions and possible TEC and TAX schemes
29
city. faction of hydro fraction of mclear —<mclear power —_fractionof _aetion of Ded
Ste of total electricity generated over + usigucd and -pemclear eftetal PV of total <Solar PV electricity
generation potealblly genasaned’ electicky electricity generation
and generated
aa a
eee total excess demand
total hess st— reassigned t
igned and x *
pei eos generated by eS staan ce ee fas generation ak
x E ° generation gencration
‘total potential fraction of gas based <zas power assigned
smuclear generation> generation of total
aon t dlectrcty gcacrated fraction of gus geactted ag _~total potetiay
ae faction of wind generated over potedtally generated generated gas power>
Lae over potentially generated
<total excess demand fraction of cleen coal es Se
reassigned to clean coal generation of total Kino! fraction oi Beeeration of total generation
generation alloc’ electricity generation ‘bipiniass: total electricity biomass of electricity generation ®
. generated over “t— eeneraed bY —M™ total
total electricity cure | ara ie, total electricity per er |
wremeratedby clan generation generated by coal and generated>
total excess ed fled
fraction of clean coal demand reassigned nati ee <total potentially
| —m generated over potential e . to biomass generated over =~ yy
focus > gentraienainc> potentially generated enerated coal power>
Figure 20: The total amount of electricity generated and the fractions of the different technologies
2.0.8 Amounts and Fractions Generated, and Monitoring and Control
The total amount of electricity generated and the fractions of the different technologies are caleu-
lated in the “Amounts and Fractions Generated’-view displayed in figure (20)
And two additional views the simulation control view and the simulation monitoring view
(see figure 21) facilitate the iterative exploration process.
References
European Commission (2006, March). EU-25 Energy Fiches. TREN C1 internet. {7| [10] {121 (17
220) 21) 23)
International Energy Agency (2000). Experience curves for energy technology policy.
IEA/OECD. 9]
International Energy Agency (2003). World Energy Investment Outlook. 2003 Insights. Paris:
OECD/IEA. (9) (10) {12} [17] 20) 21)
International Energy Agency (2004). World Energy Outlook 2004. Paris: OECD/IEA.
International Energy Agency (2005a). Electricity Information 2005. Paris: OECD/IBA. {10} (12)
International Energy Agency (2005b). Renewables Information 2005. Paris: OECD/TEA. {17
(23)
Nuclear Energy Agency (2005). Nuclear Energy Data 2005. Paris: OECD/NEA. NEA No 5989.
(14)
Pruyt, E. (2004). System dynamics models of electrical wind power potentiality. In J. Coyle
(Ed.), Proceedings of the 22nd Conference of the System Dynamics Society, Oxford. [5]
Ventana Systems (2000). Vensim DSS Reference Manual. Ventana Systems. [22]
ee
oe SL
poe ee eam / soos generated by
aioe | ach
ree a cares, /
me ee pile mentite neice
ee Siete fo wien lOrpoe metal milan aos
crete enn tet emacs yay grad el oad
ue
< Sete
ate sioNnenen aude mone eae ott
ne eee SS SS SS ar
‘tora anne anount of COP ccomminine ancant of CO2 <spectic CO2 emissions <CO2 emissions elec ~percemage pile.
oidedbr id power “ssodedby nid power felety seaated> seated by > tite capac»
total CO2 emissions as percentage mnt COD <CO2 emissions cle adeno “total oss non
Sa
TGC demand “TGC supply> average pave kTGC> eee “<eas use electricity -scoal use electricity
pls pct
octoncfgnsbosed —factonfcoal based <Fstoncfmelarce <Faconaftionse of incon dean scion of BV ofttal fakin of tycko tt
fomrainactiot peerinon oft ciemny total elect emeraion ftom ceencty, “SESS etuiee ectily pacaicn
eecasco> penersion= ‘sverain — semen
Sovecmect dean a malate me
oa power reer |< sc P
a 7 pact a
sey nome z avesmers 2006-3101 aye
aveter wid
subsidies wind pow:
Figure 21: The simulation monitoring view
30
