Table of Contents
The transition from fossil fuelled to a renewable power supply
in a deregulated electricity market
Klaus-Ole Vogstad!, Audun Botterud!, Karl Magnus Maribu!, Stine Grenaa
Jensen’
1) Norwegian University of Science and Technology (NTNU)
2) Riso National Laboratory, Denmark
phone +47 73597644 fax+47 73597250
klausv@ stud.ntnu.no
Abstract
In this paper, we investigate the trade-offs between long-term and short-term effects of
energy planning within the context of a deregulated power market. The purpose is to find
efficient policies that can aid the transition from a fossil fuelled to a renewable based pow-
ersupply. Ourcase study is on the Nordpool power market. The model focus on the main
feedback loops that determine long-term development for new capacity, namely the unit
commitment (operational characteristics), capacity acquisition, technological progress
and finally resource depletion. We show that the operational characteristcs sometimes are
important to include in long-term analyses also in long-term analyses. Finally, some sim-
ulation runs for two possible policies are presented and discussed.
Introduction
While deregulation sweeps across the electricity sector, another important change is tak-
ing place - the transition towards a renewable energy supply.
The roles of renewables play a prominent role in all the Nordic countries’ stated energy
plans. Our hydro, wind and biomass resources are plentiful, and the availability of these
resources played a crucial role for industrialising the Nordic countries. In Denmark, wind
energy revived during the energy crisis in the 70ies, and is now the 3rd largest export in-
dustry. Hydropower in Norway gave rise to energy intensive industry (Hydro, Elkem).
The paper and pulp industry in Finland and Sweden make extensive use of bio energy re-
sources. Nuclear power came into use in Sweden and Finland, but was stopped in Den-
mark and Norway. Denmark relies heavily on fossil fuels, but their Energy 21 plan aims
at phasing out fossil fuels, converting to a renewable based energy supply within 2050
(Energy 21). Sweden also formulated similar targets for a long-term sustainable energy
supply (SOU,2001). The present situation of the Nordic power supply is summarised in
Figure 1.
The Nordic electricity market (NORDPOOL) was first introduced in Norway 1991, then
expanded with Sweden in 1996, and does now include Norway, Sweden, Finland and
Denmark. The power market was initially established to improve the lack of economic ef-
ficiency in the power electricity sector, as pointed out by economists (Ferland, 1976).
Furthermore, large benefits could be obtained by better coordination of the production ca-
pacity between countries. Harmonization of tariff structures, taxes, and energy policies is
acontinuation of the deregulation process, as well as new regulatory market based mech-
anisms (i.e. CO2-quota markets, Tradable green certificate markets, etc) that can replace
the former tax/regulation policies.
In this paper, we will examine the long-term transition from fossil fuels to a renewable
NOR SWE DEN FIN Total
Supply 1999 2010 1999 2010 1999 2010 1999 2010 1999 2010
Hydro [TWh] 115 63 14.5 192.5
Wind P [TWh] - 3 - 4 3.5 8 - 1 3.5 16
Nuclear[MW] 9450 8850 2610 3810 12060 12660
CHP central [MW] 1280 570 4800 5220 2500 2750 8580 8540
CHP district [MW] 980 1916 2100 1590 730 2100 3810 5606
CHP ind [MW] 840 820 1550 1750 2390 2570
Condense [MW] 0 400 435° - 2400 0 3760 6595 400
Gas turb.[MW] 195 70 1450 1715
Demand [TWh/y] 120 123° «1430 «152 34 37 73 85 370 397
Figure 1 Installed capacity in the Nordic countries, 1999. Scenario 2010 according
to political targets in accordance with each country’s energy plans. (Source:
Vogstad et al., 2001)
electricity supply within the context of the Nordic power market by examining the main
feedback mechanisms that determine the development of new capacity in a deregulated
power market.
The use of models in long-term energy planning
The electricity sector makes extensive use of computer models in most of their activities.
Figure 3 shows some examples of the models presently in use distributed along a time
scale. At left, component design and stability analysis heavily depend on dynamic simu-
lation tools such as Simpow (not to be confused with Powersim). In production schedul-
ing and power trading, a number of short-term optimisation tools are used. For long-term
energy planning, techno-economic partial equilibrium models are used, such as Markal
and Nordmod-T. each tool only addresses a few of the feedback mechanisms relevant for
long-term energy planning, because the models involve optimisation methods that require
simplified mathematical representations. For this reason, these long-term models usually
omit some important feedback mechanisms.
Such relationships are then left to the decision maker’ policymakerss own personal judge-
ment, and too often, this is how the controversies arise. For instance, policymakerse em-
phasize short-term effects and long-term effects differently. For instance, some argue that
substitution from coal to gas is the more cost-effective environmentally sound policy
whereas others argue that renewables might be more costly in the short run, but will be
Figure 2 The Nordpool power market area: Norway, Sweden, Finland and
Denmark. Transmission lines to Russia, Poland and Germany
more cost-effective in the long run, when the learning curve effect is brought into the
equation.
Using the system dynamics approach, we try to capture the main feedback mechanisms
we believe are important for our problem (and that often cause controversies), to find ef-
ficient policies to support the transition from a fossil fuelled towards a renewable based
electricity supply .
System dynamics approach
Resource depletion
Technological progress
Regulating power Investment costs
Feedback mechanisms:
Power stability analysis
Problem: Component/Control Unit Goria
design
Energy policy
Capacity acquisition
Simulation tool: EMTDC PSSE, Simpow
Time scale: ms s min 15min, hour week season yr >20 yr
Figure 3 Energy models for decision support distributed along the time scale.
Simulation tools (blue row), Problem for decision support (red rows) and feedback
mechanisms included in our dynamic power market simulation model (green rows).
The power market model
Our model is a system dynamic representation of the Nordic electricity market with em-
phasis on the supply of various competing generation technologies. Power generation
technologies consist of the four main technologies hydropower, wind power, biomass and
thermal power. Thermal power consists of nuclear, coal, gas and peak load units (usually
gas turbines). These technologies possess different economical, technological and envi-
ronmental characteristics in terms of investment and operational costs, operational char-
acteristics, emissions, resource potential and potential for technological progress.
The common Nordic Power Market! settles the market spot price for each hour, which is
the most important information for decision makers on the supply. Additional market
based services could be the TGC market and the CO2-quota market, plus the already ex-
isting futures market where long-term power contracts are made, and the power balance
market.
Finally, the availability of resources ultimately limits the development of each technolo-
gy. The potentially available resources for each technology are described through the re-
source availability sector.
Constraints on transmission capacity between the various regions are not considered in
this model. Transmission constraints will for sure give rise to stronger price variations,
thus imposing transmission constraints will tend to amplify the mechanisms caused by the
feedback loops in Figure 4.
The time horizon is long enough for long-term impacts to take effect, while time resolu-
tion should be sufficiently small to capture the short-term mechanisms that we would
chose to include. For this reason, we have chosen a 30-year time horizon, allowing the
resource availability and technological progress of energy technologies make an impact.
Time resolution must be sufficiently small for electricity prices to adjust the demand/sup-
ply balance over the year. By doing this, we are able to simulate the capacity factor (uti-
lisation time) for each generation technology, because it is important for the profitability
and hence new investments in capacity. Wind power and hydropower will generate pow-
er even at low spot prices, while fossil fuels are characterised by their fuel costs. There-
fore, the share and seasonal variation will determine how much of the capacity is utilised
during a year.
Our focus is on the supply side of the Power Market, and the demand side is therefore less
detailed. However, we try to capture some of the characteristics that are of importance
for price formation: Underlying demand growth and price elasticity of demand. Different
developments in demand can be assessed by sensitivity analysis. In fact, there are few
strong feedback mechanisms between the demand side and the supply for electricity.
1. For more infomation about the organisation of Nordpool, see www.nordpool.no
Feedback loops
On the supply side, several feedback loops determine the development of installed capac-
ity and electricity generation for each technology (see Figure 4).
Fractional growth
rate
- | 4 balance ~ , .
Price elasticity of
Price of electrici demand
B1- Unit
Electricity commitment
generation + Capacity factor
+
B4- erosion of operational costs
Resource CF
availability Capacity B2- Capacity
he } acquisition
+ +
Technological a
progress R1- Learning Expected profitability
curve of new capacity
B3 - Resource - Investment &
depletion operational costs
Figure 4 Main loops of the electricity supply side
Unit commitment (B1) is the process of operating units hour by hour to serve current de-
mand loads. Generating units with the lower operational costs are commissioned first and
the units with the highest marginal costs are the last units to be commissionedned. The
last unit in operation determine the spot price at each time point.
Capacity acquisition (B2) is the process of new capacity investments based on the expect-
ed of profitability of new capacity additions. Long time delays are involved in this proc-
ess, because applications must be sent to the regulating authorities, before new
developments can be made. This process could take several years, so price expectations
are based on forecasts several years ahead. The process of developing new capacity varies
depending on technology type. Expansion from hydropower typically is a tedious proc-
ess, because a large number of stakeholders are involved. Less time delays are involved
in wind power and biomass, with usually fewer stakeholders.
The learning curve effect (R1) is a reinforcing loop, which is more prominent for the wind
power and other new renewables, than for mature technologies such as fossil fuels and hy-
dropower, although improvements are also made within these technologies. Moxnes
(1992) shows how positive feedback loops of learning curves can be used in policymaking
Resource depletion (B3) is the ultimate limiting loop. Potentials for large-scale hydro-
power are almost exhausted in the Nordic countries, while it is assumed that the availabil-
ity of fossil resources does not constrain thermal energy generation within the time
horizon of our model, due to the large gas resources in Russia and Norway. On the other
hand, availability of windy areas do however constrain the development of onshore wind
power- but offshore potentials for wind power provides new (yet more expensive) oppor-
tunities.
Erosion of capacity factor (B4) is one possible mechanism that could speed up the down-
sizing of thermal generation. We will address this issue later on in the model description
of each technology (cross-reference to that section). If the capacity factor (that is, the uti-
lisation of capacity) remains low, that capacity will most likely be retired before the end
of its economical lifetime. Hence, the capacity factor influence the construction and re-
tirements of thermal capacity - which was the motivation for including the unit commit-
ment loop in our long-term simulation model.
In the following section, we will give a more detailed description of each sector in Figure
5, starting with the capacity acquisition of each generation technology. The layers of
power generation technology illustrates that there is one structure for each type of tech-
nology, and there are some structural differences between these.
Capacity acquisition
Figure 6 shows the process of submitting an application to develop new capacity, where-
upon the application is processed. A pproved applications can then be developed into new
capacity that comes on line. This process is highly regulated and involves long time de-
lays, depending on technology type. For instance, the process of developing new hydro-
power plants could take several years, because of all the stakeholders involved (NGO’s,
local authorities, national authorities. The final decisions are often made by the parlia-
ment. Also, thermal generation such as nuclear, coal - and even gas power catches public
and political debate in each of the Nordic countries. A final decision is now being made
regarding expansion of nuclear power in Finland; Sweden are discussing how fast the nu-
clear power should be faced out, as the parliament decided to phase out their nuclear pow-
er. In Denmark, no new coal-fired plants are approved, and areas for onshore wind power
are scarce. One Norwegian government resigned as they refused approving the first land-
based Norwegian gas power plant!. By the time applications finally are approved, the
profitability of the project may have changed, or new environmental requirements have
made the projects less attractive. The reject fraction and projects abandoned are exoge-
nously determined.
1. The Norwegian power supply is 100% based on hydropower, except the offshore installations,
which is supplied by gas power.
(7 \
Power generation technologies
Power Market
* Spot price
+ Hydropower + Futures market!
* Wind Power + TGC market
+ Biomass CO2-quota market!
* Thermal power
- Nuclear Unit commitment 1) Not implemented yet
eo * Capacity Factor
° * Marginal producti
- Peak load gt eae procuenon
General characteristics
apacity acquisition
* Applications
* Applications approved
* Capacity on line Demand
* Exogenous growth
rate
* Price elasticity of
demand
Profitability assessment
* Net present value calculations
Price forecasts
Investment costs
Operational costs
Tax/subsidy intervention
Resource availability
Wind area, Remaining hydro re-
sources, Biomass, Gas reserves
N
XN S
Figure 5 Structure of the model, subdivided into sectors.
The initial values, starting from 2000, are set so that the supply line is in dynamic equi-
librium, that is, the initial application rate balances the discard rate when the effect of
profitability on application rate is 1 (see definition of Effect of profitability on application
rate in the Capacity acquisition section). Construction time and application processing
time typically varies from technology to technology, the shortest are for biomass and wind
power, (about 1/2 a year), while the longest time delays are for hydropower (3 years ap-
plication processing time, 3 years construction time)
Reinvestments in existing capacity are not included in our model, although it might be
necessary to include the vintage structure of the rapidly developing wind power technol-
(ec apacity acquisition~~ ~~
y SU |e
apni \ c
fetime
S
Effoct of profitability Resource usaue rate
‘on applicaton rato
Figure 6C apacity acquisition, taking wind power as an example.
ogy. The technological performance of this fast growing technology changes significant-
ly over the time period, and new turbines require much less area compared with older
ones.
Profitability assessment
The variables involved in the profitability assessment are displayed in Figure 8. Profita-
bility of new capacity investments depends on the one hand on expected future spot pric-
es, and on the other hand on investment costs, operational costs and expected capacity
utilisation. The learning curve reduces investment costs, and operational costs.
Energy investment costs = Investment costs * learning curve / (lifetime * max full load
hrs * CF) [NOK/MWh]
Profitability indicator = (Expected electricity price - operational costs) / Energy invest-
ment costs. [NOK/MWh]/[NOK/MWh]
The profitability indicator is held against the required return on investment for that tech-
nology, where we used 7% interest rate and 20,30 and 40 years expected lifetime of wind,
natural gas and hydropower respectively. Required return on investment and profitabil-
ity indicator has the same interpretation as the annuity factor. Finally, these two factors
are used in the “Effect of profitability on application rate” :
Effect of profitability on application rate = graph(Profitability indicator /Required retum
on investment)
There is surprisingly little theory or information to obtain from standard economic litera-
ture on this relationship. Morthorst (1999) made an empirical estimation of this relation-
ship on wind turbine owners in Denmark, and came up with an s-shaped curve for private-
owned turbines. There was no similar study for other generation technologies available.
Application rate wp Expansion rate Capacity WP
id l
operational costs
oe
aie
oe
Figure 7Expected profitability of new capacity. Profitability depends on spot price,
expected capacity utilisation, subsidies and the learning curve. Costs are divided
into investment costs and operational costs. Market prices are forecasted based on
previous values.
Effect of profitability on application rate and investment rate
Effect of profitability on application rate
6
b
Profitability indicatorRequired return on investment [Dimensionless]
Figure 8 The relationship between expected profitability and application &
investment rate
We can, however derive the shape of the curve analytically, based on some simple as-
sumptions, see Figure 8. The relationship would represent a cumulative lognormal or nor-
mal distributed curve as the sum of the uncertain factors involved in the profitability
assessment. The curve is applied on all technologies. The magnitude of this curve is lim-
ited upward so that the maximum application & investment rate is set to 30 % of installed
capacity. Rapid growth within an industry is constrained by the availability of goods and
services from suppliers in other sectors.
Unit commitment
The operational characteristics differ between the technologies. Thermal units are oper-
ated after their marginal costs of generation, which is roughly equal to the fuel costs.
Thermal plants take spot price as an input to determine their generation level. As the pric-
es of electricity rise, plants with increasingly higher marginal operational costs starts up
until the spot price levels out.
nit commitment
C02 Emissions Gas
ae
eo __
Capacity Gas ‘coi
voctuticse nse
Narginal co2-
= emissions per MWh
oo
enratin th
/
x
/
go
e
Expected CF gas
Figure 9Unit commitment. Capacity utilisation expressed by the dimensionless
capacity factor is a direct function of market price.
I
Utilities must decide when and how much of their capacity should be in operation. This
is known as the unit commitment problem. Thermal units are run by their marginal fuel
costs. We denote the fraction of capacity in operation as the Capacity Factor (CF). The
capacity factor can then be calculated using the marginal cost curve of thermal energy,
normalised by installed capacity. We thus assume the shape of this curve (that is, the dis-
tribution of baseload, medium load and peak load) to be fairly constant over the simula-
tion time. At each point in time, electricity prices determine the fraction of installed
capacity in operation. The unit commitment submodel is shown for gas power in Figure
9, and this structure applies to the thermal generation types (including biomass). The
graphs in Figure 10 are based on V ogstad et al (2001). In the system dynamic model, the
CF curves were somewhat smoothed in comparison to the the original data because the
original data are aggregated.
Wind power generation cannot be controlled, and must generate power when the wind
blows, although wind power can be taken into account in hydropwer scheduling (V ogstad,
x aot Marginal costs of generation, thermal units
Cumulative thermal capacity [MW]
080100 1300030300380 a0 aS
Marginal costs of generation [NOK/MWh] Marginal costs of generation [NOK/MWh]
Figure 10 Marginal cost curves, thermal power, Nordpool. Left : Aggregate cost
curve. Right: Marginal cost curves for each technology type.
2000) This is also the case for run-of-river hydropower. Figure 11 shows the variation in
yearly hydro inflow and wind power. The right graph shows the seasonal variations in
demand, hydro inflow and wind power. While wind power nicely fits the seasonal de-
mand curve, most hydro inflow is released during the spring.. From 30-60 years of inflow
data with resolution of one week, we represent the yearly variations for wind power and
hydropower as normal distributed random variables with a standard deviation of 0.063
and 0.12 respectively. Our simulation runs therefore includes some stochasticity, but all
the simulation runs have the same stochastic series for ease of comparison. Hydro inflow,
wind and demand are in addition represented with seasonal variations using the profile
curves in the rightmost graph of Figure 11.
/ Stain
Ftp tse] —
(__Eoimne
4
ws [Sn jfin-
Wind power
a —Hyriee power
Normalised annual production (94)
Normalised weekly data @)
|
3
[-——~—
i 0
1961 1965 1969 197319771921 1985 1980 1 4 7 101316 192225 28 31 34 37 40 43 46 4952
Yer
‘Week of year
Figure 11L eft: Yearly variations of hydro inflow and wind energy. Right: Seasonal
variations in hydro inflow, wind energy and demand. (Source: Tande & Vogstad,
1999)
Run-of-river hydropower is very similar to the intermittent wind power in the sense that
neither of these units can be scheduled for generation.
Hydropower with reservoirs, however, has the unique possibility of storing energy. The
reservoir capacity is not unlimited, therefore “marginal costs” of hydropower generation
is associated with expected profits of storing the water for later use. Expected future prof-
its that can be obtained by storing water for later usage depend on future hydro inflow and
electricity prices, and present reservoir level. The method of calculating the “marginal
costs” of hydropower generation is referred to as the water value method, and utilities use
sophisticated optimisation models to accomplish this task!. We approximate this hydro-
power scheduling problem using a table lookup function of water values taking the res-
ervoir level content as input, see Figure 13, left graph. Figure 12 illustrates the
hydropower scheduling problem and the management of reservoirs
Atmaximum reservoir level, the water value is 0, because you will not be able to store
(— Unit commitment hp ™~
hydro inflow ® Ag fl cas hr run
i \ @
/ Ou:
e.
Generation
Capacity hp
lage
pillage time
Max reservoir level
e; @ x) Nargear
us Le ater value ~ S/S
e-
yearly average price
Figure 12Production scheduling (unit commitment) for hydropower with reservoirs
new hydro inflow, and the water is spilled. When the reservoir level is about to run empty,
prices would rise, because there is now a scarcity of supply in the system.
The Nordpool Power Market
30 % of financial power contracts were settled through the Nordpool Power Exchange in
2001. The rest is traded through bilateral contracts. Market price clears every hour, and
anumber of other financial derivatives are offered, for instance the regulating power mar-
ket (to balance power) and futures contracts for power up to 3 years ahead.
Price is an important input to all other sectors; unit commitment, profitability assessment
and demand.
The spot price is calculated as follows:
(Demand, _ , - Generation, _ D 1
1 Demand, _ ; “Time to adjust price INOK/MWh (1)
Price, = Price, .
1. The energy models EOPS and EMPS mentioned in Figure 3 are actually used for this purpose
Water value (marginal costs) for hydropower generation rere
4
Normalised water value [dimensionless]
Normalised reservoir level (dimensionless)
1 reservcir ive (ration)
Figure 13Left: Water value column, representing water values as a function of the
reservoir level. Right: Water values calculated using EMPS. The water value
column varies over the season, but not dramatically.
Generation hp
Runt er e generation co
eo” Generation bio” Generation nu
a =, saveration gs
~ wee ane th
2. pl
Time tp adjust price
| ‘Qe.
yearly average price
Figure 14 A dynamic formulation of price setting in the Nordpool Power Market
This represents a hill-climbing search to find the price that balances the supply/demand.
The power market has some special features. The bids of demand and supply are submit-
ted 12-36 hours in advance of market clearing. Suppliers make their bids using a supply
curve. Nordpool then collects all bids and determine spot price on the basis of the aggre-
gated curve. Market actors are then well informed about how much they are obliged to
supply some 12-36 hours in advance of market clearing. If utilities fail to fulfil their con-
tract obligations, they will be charged after hand by the power balance market. The power
balance market is designed to provide generation on short notice, to adjust for imbalances
and unexpected failures. The power balance market price is usually higher than the spot
market, although hydropower in Norway provides a fast and cheap way to up/down reg-
ulate power. In thermal dominated systems, power for rapid up/down regulation is more
expensive.
Market participants are assumed to behave boundedly rational when making decisions
about new capacity investments (see the Capacity acquisition and the profitability assess-
ment submodel). Currently, we only include the spot market in the Power market sub-
model. Investors will typically use the futures market to estimate long-term power prices.
If we wanted to study consequences of strategic behaviour in these markets, we could ex-
tend our model with the power balance market and the futures market. Prior to that, a mar-
ket for Tradable Green Certificates, and a CO2-quota market will be implemented.
At present, there is no strategic behaviour of market participants, but this can be easily
implemented with the dynamic representation of the spot market.
In our model, price is adjusted every 3rd day. It turns out that market balances sufficiently
for our purpose, which is to give an adequate representation of the operational character-
istics for the various generation types. As a result, the capacity factor CF changes due to
seasonal variations of wind power and hydropower, and the interplay between thermal
units with higher marginal operational costs.
Learning curves
Each technology has reached different stages of maturity. Technology progresses as the
cumulative installed capacity of a technology increase, but we must keep in mind that the
Nordic electricity supply is only a small fraction of the total world market for electricity,
and technological progress cannot be endogenously described by the cumulative installed
capacities in the Nordpool area. We therefore define an exogenous leaming curve mul-
tiplier and a learning rate, k, of technology i.
Leaming curve multiplier(t) = g Be 4) (dimensionless) (2)
Investment costs,(t) = Initial investment cost, * Learning curve multiplier,(t)
Technology ; Hydro Wind Natural gas Bio energy
k, 0.002 0.014 0.05 0,008
Figure 15 Exogenous learning rate for the various technologies.
However, we take wind power as an exception. The major share of wind power develop-
ment has taken place in Denmark, and Danish wind turbine manufacturers dominate the
world market by 50% share. Denmark is now focusing on the offshore wind power de-
velopment, and several large-scale offshore farm projects are under construction. The
shallow waters around Denmark and Sweden are especially suited for the offshore wind
power technology, and we assume new technological progress to be closely related to the
wind power development in the Nordic countries introducing the following standard
function (IEA, 2000):
i -0.2
Nordic market learning curve multiplier Geese we
The large share of bio energy and hydropower could also justify an exogenous learning
curve for these technologies, but we leave this issue for later work.
Resource depletion
The four technologies will ultimately reach limits of resource availability. large-scale hy-
dropower has in practice reached its potential, though these limits are to some extent po-
litically and economically determined. Further expansions must come from small-scale
development.
._ Remaining potentials forhydropowerin Norway (Source: NVE)
esource depletion hydropower
L@ *¢ fut 1004 hours
”. (3
Energy costs INOK/MWh]
o $ 10 a %
Resources [TWhit]
Figure 16C osts of new hydropower development in Norway. (Source : NVE)
Wind power has still a large, unexploited potential, world wide as well as in the Nordic
countries. However, Denmark is about to reach its limits for land resources. Further wind
power developments are taking place offshore, or by upgrading old farms onshore. When
examining the vintage structure of wind power onshore, it seems possible to increase in-
stalled capacity from 2500 to 6000 MW by substituting old turbines with the new and
larger turbines on existing sites. At present, the energy costs of developing offshore parks
are at least 30% larger than for onshore parks. In contrast, the offshore potential is prac-
tically unlimited. The coastal areas of Norway provide good opportunities for cheap and
cost-effective wind power onshore, and a recent wind resource assessment has been made
by NVE!. The resource availability influence costs of wind power in the same way as with
hydropower (see Figure 16), but differ in the sense that wind resources are measured in
square km’s of windy areas. The remaining potential in turn determines the capacity fac-
tor for wind. As the more windy areas are utilised, less windy areas remain to be devel-
oped at higher costs.
1. See Vector: www.vector.no
Due to the large gas resources in Norway and Russia, we assume natural gas generation
not to be restricted by resource availability during the next 30 years. Rather, environmen-
tal concerns put restrictions on emission levels and resource availability of natural gas or
other fossil resources are not estimated in our model. A further
Scandinavia is largely covered by pine and spruce trees, with a large pulp industry. Dif-
ferent kinds of waste from the pulp industry, building materials from wood, etc, provide
cheap sources of biomass for heat and electricity cogeneration. The available potential is
estimated to be about 220 TWh/yr (for both energy and heating), but the costs of the
sources limit the economical potential. Waste from pulp industry and other waste mate-
rials from wood are the cheapest, direct use of wood for heating a bit more expensive, and
finally growing energy crops is the most expensive alternative. The costs of biomass
therefore increase as more expensive sources for biomass must be utilised.
Export
Total transmission capacity for exports amounts to 3500 MW, and can be regarded both
as supply and demand. The profitability of transmission lines depends on the price dif-
ferences between Nordpool and neighbouring countries. Deregulation and restructuring
of the electricity sector has stopped utilities from investing in new transmission capacity.
One the one hand, if the price difference between Nordpool and neighbouring countries
is high, the transmission itself is profitable. On the other hand - new transmission cables
will also reduce the spot prices in the Nordpool area, because the capacity not being used
is available to all market participants by the Nordpool. Power intensive consumers could,
however profit on building new transmission lines. A study made by Wangensteen et al
(1999), concluded that it is probably not profitable for utilities to build new transmission
lines under the current circumstances. Exchange capacity is therefore fixed in our ap-
proach.
xchange
Marginal price
ag
\
Price rag
y
J
6 Exchange capacity > ke
Hew power cables Cables discard rate Price rato [dimensionless]
Capacity utilisation
Capacity factor dimensionless]
Figure 17Transmission lines from the Nordpool area to neigbouring countries.
Right: Capacity utilisation as a function of the price ratio
Demand side
Our focus is on the supply side, though there are many interesting alternative options to
new developments of capacity on the supply side. As a compromise, we try to capture
some of the main characteristics of the demand side: Demand variations (seasonal), de-
mand growth, and price elasticity of demand. Changes in demand consist of a net frac-
tional growth rate (due to increase in population, income, etc), and a fractional price
elasticity of demand. Consumers compare present end-user prices with a reference price,
which is an exponential smoothing of last 5 years’ end-user electricity prices. There is no
distinction between different types of consumers.
(— Demand Seasonal demand Domai leh >
seasonal variation
variation
Fractional demand
growth
Wet-Growith rate
knesbarnielnetet Initial demand
oe fdemand effect of price on
demand growth
Energy dependent Price normal
grid tari
Perceived end user
red ait ms
price
LL fd user price yw,
2 é
Price
Fractional change in
price
Price change
Figure 18 Demand submodel. Underlying growth and price elasticity of demand
represent end-user behaviour. The end-user slowly adapts to new prices
Simulation results: Base run
In our base run, a subsidy of 70 NOK/MWh is paid to new renewables (wind power and
biomass). Figure 19 shows the development of installed capacity, generation and prices/
costs. While nuclear power and coal power are decomissioned, wind power and gas pow-
er picks up around 2015. Bio energy shows a faster development, but peaks around 2015.
Although the potential for biomass is large, the availability of cheap biomass is more re-
stricted. Waste residuals from building materials and the pulping industry are cheap
sources of biomass, but when this potential is exhausted, more expensive sources must be
used, either direct extraction from the forest, or growing of energy crops. Biomass is also
affected by the capacity utilisiation, which is a function of the price level. Even though
technological progress improves the efficiency of biomass, other sources are more com-
petitive in the long run of bringing the costs down.
Price increase to a higher level during the first five years. As mentioned earlier, we as-
sume that no new nuclear or coal power is built, only gas power is from an environmental
point of view accepted among the fossil technologies. The long-term energy costs of gas
power were initially higher than the market spot price. Thermal power dictates the elec-
tricity price, and when the prices increase, the capacity utilisation must also increase.
There is still overcapacity in the Nordpool market and we are still in the transition phase
towards a power supply with a market price that will be in equilibrium with the long-term
marginal costs of generation. A first insight with our model - we can expect the Nordic
market prices to increase to a new long-term level equal to the long-term marginal costs
of developing new gas power within this scenario.
We also observe that the total costs of wind power fall to a minimum before increasing
(total costs are displayed with subsidies included). The learning curve effect drives the
costs down, but as the most attractive sites are developed first, more costly and remote
areas must be developed, which will increase costs in the long run. The wind power de-
velopment is therefore likely to follow an s-shaped trajectory of capacity development. If
sufficient technological progress is made for offshore wind power, potential areas are no
longer a constraint. Other constraints, perhaps grid constraints would then be limiting fac-
tors.
The simulations in Figure 20 show the base run when variations in hydro inflow, wind
energy and demand is included. As can be seen, omitting seasonal variations changes the
simulation results to some extent. Simulations without seasonal variation are shown as
thin lines in the same graphs. The installed capacity of wind power, hydropower and even
biomass increase at the expense of gas power. In fact, the CF and expected CF are re-
duced due to these seasonal variations as shown in Figure 21.
The below Figure 21 shows how the supply and demand balances for each simulation run.
The smooth lines correspond to the base run without seasonal variations. The oscillating
lines show demand and supply when seasonal variations are present.
Figure 22 Is a close-up of the generation and price graphs in Figure 19, respectively. The
lower graph displays the simulated spot price, and also includes the observed spot prices
for 2000 and 2001 for comparison. 2000 was a wet year, and therefore the prices were
notably lower during the first months of the year. The corresponding CO2-emissions are
shown in Figure 23, both with and without seasonal variations. The CO2 emission levels,
and even the behaviour, changes when seasonal variations are included due to the nonlin-
ear operational characteristics of the unit commitment loop (see Figure 4). First, we
should note the substitution from coal to gas, which will reduce the CO2-emissions for
some time according to the upper brown thin line in Figure 23. However, nuclear power
also needs to be replaced, and in the base run scenario, there will be a net increase in CO2-
emissions after 2015. Therefore, increased efforts of increasing the share of renewables
are required if the power industry is to meet the Kyoto target for greenhouse gas emis-
sions.
Installed Capacity
mW, — “Capacity Gas
- Capacity Gas
aaiea0 ~ *Capacity Bio
— Capacity Bio
— *Capacity hp
30,000: — Capacity hp
~*Capacity NuP
20,000: — capacity NuP
~ *Capacity co
10,000 ~ Capacity ca
—* Capacity WP
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030 [> Capacity WP
General
Twhiyr
=*Generation wp
- Generation wp
—*generation hp
generation hp
—*generation co
generation co
~* Exchange
Exchange
50. —*Generation Bio
Generation Bio
~*Generation Gas
0 Generation Gas
01/01/2000 01/01/2005 01/01/2010 01/01/2018 01/01/2020 01/01/2025 04/01/2030
150:
100:
Price, costs
NOk/MWh,
=*Price
= Price
—*Total eneray costs wp
Total energy costs wp
~ "Total Energy Costs Bio
Total Energy Costs Bio
—*Total eneray costs gas
150:
Total energy costs gas
200:
01/01/2000 01/03/2008 01/03/2010 01/01/2015 01/01/2020 01/01/2025 03/01/2030
Figure 19 Base run simulation. Subsidy for wind power and biomass: 70 NO K/
MWh. Demand side as described in previous section.
Installed Capacity
baa — *Capacity Gas
50,000 — Capacity Gas
~*Capacity Bio
40,000 — Capacity Bio
~ "Capacity hp
30,000 — Capacity hp
—*Capacity NuP
20,000 = Capacity NuP
~ *Capacity co
10,000: ~ Capacity co
— *Capacity WP
01/01/2000 01/01/2008 03/01/2010 04/01/2018 01/01/2020 01/03/2025 01/01/2030 | CaPacty WP
Generation
|—*Generation wp
— Generation wp
~"generation hp
AN | [eon
| \ | —* generation co
generation co
\ —*Exchange
ALAM exchange
¥ —*Generation Bio
|— Generation Bio
—*Generation Gas
Generation Gas
o1/o1/2000 01/01/2005 ai/oi/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030
Price, costs
NOK/MWh
800-
—*Price
600- Price
—*Total energy costs wp
Total energy costs wp
400- 1 |—*Total Energy Costs Bio
Total Energy Costs Bio
I Woy HiT |-*Total eneray costs gas,
200. iT A it Total eneray costs gas
yov ¥
01/04/2000 01/01/2008 01/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030
Figure 20 Base run simulation with seasonal variation in demand, hydro inflow
and wind energy.
sd tl j if WU UU a (1),
| | | —CF gas
06 | —*CF gas
— Expected CF ges
— “Expected CF aes
03
01/01/2000 01/01/2010 01/01/2020 01/01/2030
Market equilibrium chec
Twhiyr
*Demand
= *seneration
Demand
— generation
o+
o1/01/2000 01/01/2010) 01/01/2020 01/01/2030
Figure 21 Capacity factor gas power. Blue lines show simulations without seasonal
variations, red lines show C F when seasonal variations are included in the
simulation.
Generation
Twhiyr
— *Generation wp
— Generation wp
— *generation hp
= generation hp.
— *2un-of-river generation
— Run-of-river generation
— "generation co
— generation co
— *exchange
— Exchange
— *Generation Bio
— Generation Bio
01/01/2000 01/01/2001 01/01/2002 01/01/2003 01/01/2004 01/01/2005
Price, costs
NOK/MWwh_
Observed
rked spot price 2000,2001
_—*
zoo Price
— Price
— *Total energy costs wp
\ | — Total energy costs wp
i 1 _ 1 — Total Energy Costs Blo
a a ee krotal energy tosts'gas
01/01/2000.” 0101/2001 Oifo1/2002 1/01/2003 01/01/2004 01/01/2005
Figure 22C lose-up of simulations. Upper graph shows the unit commitment of each
generation technology. The lower graph shows how prices develop during the first
years. For comparison, the observed Nordpool spot prices for 2001 and 2002 are
included. Year 2000 was an exceptional wet year, therefore, prices were kept
extremely low during the late winter and spring period. Year 2001 is more of an
average year of inflow.
Mtonnesyr
Pa
ao ad
— *Avg CO2 emission co
604 = Avg CO2 emission co
*avg CO2 emission gas
= Avg CO2 emission gas
ao — *avg COZ emission pl
= Ava CO2 emission pl
— *Tatal CO2 emissions
oo = Total C02 emissions
0 = Aaa : J \ —
1/01/2000 01/01/2010 01/01/2020 01/01/2030
Figure 23C 02 emissions,base run. Without seasonal variations: Thin lines. With
seasonal variations: Thick lines.
Policies for change
The most obvious policy instrument is tax or subsidy interventions. ExtemE, a 10-year
EU-study assessed the external costs of energy technologies, using risk environmental as-
sessment as a method. The cost estimates include effects of air pollution, occupational
disease, accidents and damage on natural and build environment. They used a spatial
model for emissions, because damage of for instance SO2 and NOx depend on the loca-
tion. Figure 24 summarises external costs of generation in the Nordic countries. Con-
verted to Norwegian currencies, the taxes for Natural gas should be in the range of 60-
640 NOK/MWh to compensate for external costs.
Extemal costs of generation values in Den Nor Swe Fin
mECU/KWh
Natural gas 7.1-80 7.7-19.2 18-42 (coal) 4.2-9.6 (coal)
Bio energy 16-4.4 24 18 6-14
Wind 0.6-3.65 0.5-1.1
Hydropower - 2.3 0.3-54
Figure 24Summary of ExternE results on external costs of energy technologies.
(Source: ExternE, http://externe,jrc.es/)
The effect of introducing a CO2-tax
In this next simulation run, we have imposed a CO2 tax from 2005 on, keeping subsidies
for renewables at the same level. The results are shown in Figure 25 and Figure 26, where
the base run scenario is shown with thin lines. We observe a development in favour of
renewables, both in terms of installed capacity and in terms of generation, as was expect-
ed. Comparing Figure 26 with Figure 25, including seasonal variations changes the re-
sults in favour of more renewables. Higher price volatility is to be expected, but not
significantly higher than that of the base run. Figure 27 shows how CO2-emissions drop
as aresult. Thin lines display base run without seasonal variations, thick lines display re-
sults with seasonal variations. The spot price will increase to a level between 200 and
250 NOK/MWh, which is somewhat higher than in the base run.
Installed Capacity
HW. — "Capacity Gas
50,009 — Capacity Gas
~*Capacity Bio
40,000: |~ Capacity Bio
|— “Capacity hp
30,000 | capacity hp
~*Capacity NuP
20,000 | capacity NuP
~ *Capacity co
10,000: ~ Capacity co
— "Capacity WP
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 01/01/2028 01/03/2020 | C@P Cty WP
Generation
Twhiyr
200: =*Generation wp
Generation wp
—*generation hp
150: generation hp
—* generation co
generation co
~*Exchange
Exchange
50. —*Generation Bio
Generation Bio
~*Generation Gas
0 Generation Gas
ox/o1/2000 02/01/2005 02/01/2010 02/01/2015 01/01/2020 03/01/2025 01/01/2030
100:
Price, costs
NOK/MWh_
= *Price
|= Price
—"Total energy costs wp
|—Total eneray costs wo
200 ~ “Total Eneray Costs Bio
Total Eneray Costs Bio
—*Total eneray costs gas
4150. [Total eneray costs gas
250:
91/01/2000 01/01/2008 01/01/2010 01/01/2018 01/01/2020 01/01/2025 01/01/2030
Figure 25C 0 2-taxes of 125 NOK/MWh imposed from 2005. Thin lines display base
run values, thick line display new simulation with C 0 2-taxes. No seasonal
variation
Installed Capacity
Mw. | *Capacity Gas
‘50,000: |— Capacity Gas
~ "Capacity Bio
40,000- |~ Capacity Bio
— *Capacity hp
30,000: |— Capacity hp
— *Capacity NuP
20,000. — Capacity NuP
~ *Capacity co
10,000 | Capacity co
—*Capacity WP
|— Capacity WP.
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030
Generation
TWh/yr
— "Generation wp
— Generation wp
B00 —*generation hp
generation hp
$50) —*generation co
= generation co
100: ~ "Exchange
Exchange
—*Generation Bio
— Generation Bio
~*Generation Gas
~ Generation Gas
50:
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 1/01/2025 01/01/2030
Price, costs
Nok/MWh
=*Price
Price
en —*Total eneray costs wp
|-Total energy costs wp
—"Total Energy Costs Bio
|—Total Energy Costs Bio
—*Total energy costs gas
i Ve —Total energy costs gas
300 |
edd iim a
04/04/2000 01/01/2005 01/01/2010 01/01/2015 04/01/2020 01/01/2025 01/01/2030
Figure 26C 0 2-tax of 125 NOK/MWh imposed on base run.
Mtonne/yr
—*Ayg CO2 emission co
Avg CO2 emission co
*Aavg CO2 emission gas
Avg CO2 emission gas
—#Avg CO2 emission pl
Avg CO2 emission pl
“Total CO2 emissions
30
ee
0
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030
Figure 27Introducing a C 02-tax of 125 NOK/tonne (thick line). Thin line with
corresponding coloursdisplay base run.
The effect of a strict regulation regime
As mentioned in the capacity acquisition section, the application procedure is a tedious
process. NGO’s have been especially successful in filing complaints, which will slow
down the application processing rate significantly as prices go up. There are many envi-
ronmental laws and regulations, both in Norway and in the EU countries that would take
along time to clarify in relation to a project. This is now the case for several Norwegian
gas power plant developments, where there are uncertainties regarding obligations and
new regulations. Also, wind power projects have been exposed to this strategy, if the area
for development houses some protected species. The effect of slowing down the capacity
acquisition chain, is a reduced capacity expansion for all technologies. This is counter-
acted by the increased profitability, as prices goes up. Figure 28 that when stricter / slow-
er regulation policy has the effect of slowing down new capacity developments
comparatively less gas power is developed and wind energy generation will be higher than
electricity from gas by 2030. The prices risehydropowerhydropower to about 250 NOK/
MWh. When seasonal variations are included (Figure 29), wind power grows even
stronger and gas and biomass contribute equally to the energy mix by 2030. Even though
peak load turbines are installed to keep a fixed percentage capacity margin.
Minor changes in CO2-emissions occur, because coal power is not affected by this strat-
egy, and CO2-tax worked more efficiently in addressing the CO2-problem. Because the
margins in this system are smaller, price volatility increase significantly.
It is important to point out that NGO’s and public opinion as stakeholders have the possi-
bility of using this strategy.
Installed Capacity
HW. — *Capacity Gas
50,000: — Capacity Gas
"Capacity Bio
40,000: — Capacity Bio
"Capacity hp
30,000 | Capacity hp
~ "Capacity NUP.
20,000: | Capacity NuP
~ “Capacity co
10,000: — Capacity co
"Capacity WP
01/01/2000 04/01/2005 01/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030 [_-aPacity WP
Generation
Twh/yr
=*Generation wp
Generation wp
150: |—* generation hp.
— generation hp
—*generation co
100; = generation co
—*Exchange
Exchange
50 —*Generation Bio
Generation Bio
~* Generation Gas
o. — Generation Gas
01/01/2000 04/01/2005 01/03/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030
Price, costs
NOK/MWh
=*Price
= Price
|—*Total energy costs wp
Total energy costs wp
200 —*Tatal Eneray Costs Bio
Total Energy Costs Bio
—""Total energy costs gas
150. Total energy costs gas
250:
01/01/2000 01/01/2005 03/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2090
Figure 28T he effect of a stricter application procedure. Applications rejected: 30%
for all technologies except biomass (no reject). Application processing time increase
by 1 year for biomass and wind, and 2 years for hydropower and gas.
Installed Capacity
Mw —*Capacity Gas
50,000 Capacity Gas
~ *Capacity Bio
40,000 — capacity Bio
~*Capacity hp
30,000 - Capacity hp
— *Capacity NuP
20,000 Capacity NuP
~ *Capacity co
10,000: ~ Capacity co
— *Capacity WP
01/01/2000 01/01/2005 01/01/2010 04/01/2015 01/01/2020 01/01/2025 01/01/2020 [_CaPacty WP
Generation
TWhivr
“Generation wp
Generation wp
200 “generation hp
|= generation hp.
SD —*generation co
generation ca
100: —* Exchange:
|— Exchange
—*Generation Bio
Generation Bio
—*Generation Gas
|— Generation Gas
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 01/01/2025 01/01/2030
Price, costs
NOK/MWh
1,500
=* Prive
Price
—*Total eneray costs wp
2,000 —Total energy costs wp
~*Total Energy Costs Bio
Total Energy Costs Bio
500 -*Total energy costs gas
Total energy costs gas
o+
01/01/2000 01/01/2005 01/01/2010 01/01/2015 01/01/2020 01/03/2025 01/01/2030
Figure 29 The effect of a stricter application process. Seasonal variations included
(thick line) Thin line: no seasonal variation.
Mtonne/yr
60:
30:
C02 emissions
01/03/2000 01/01/2005 01/01/2010 o4/01/2015 o14/01/2020 o1/01/2025 01/01/2030
—*Avg CO2 emission co
|—Avg CO2 emission co
—*Avg CO2 emission gas
Avg CO2 emission gas
—*Avg CO2 emission pl
|-Avg CO2 emission pl
—*Total CO2 emissions
|- Total CO2 emissions
Figure 30 CO2 emissions as a result of stricter application process: A minor
change in emission levels. C oal power is not affected by this strategy.
Conclusion
The few policies evaluated showed to have some long-term impact. We have only com-
pared a few possible policies, and a combination of various policies could prove to be
more efficient. Some new market-based policy instruments will be implemented and
tested. These are the CO2-quota markets, and the Tradable green certificate system.
Both instruments are supposed to replace direct tax/subsidy interventions, and such a
simulation model can simulate possible problems in the design of these.
There is a need to evaluate these policies against each other in terms of efficiency, for in-
stance discounting subsidies or the socio-economic surplus for each policy and scenario
to net present value as a measure of total costs. Further research will be carried out in this
direction.
Introducing new markets increase the possibilities of strategic behaviour, that can be stud-
ied through through this system dynamic model.
References/Bibliography
Botterud, A., Korpas, M., Vogstad, K-O., Wangensteen, I., (2001): "A Dynamic
Simulation Model for Long-Term Analysis of the Power Market". Submitted to the Power
Systems Computation Conference 2002, Sevilla, Spain.
Nordpool, 2001. Elbgrsen Market Report 4/2001. Available [online] 09.04.2002 http://
www.nordpool.no/
Moxnes, E. 1992 : Positive feedback economics and the competition between hard and
soft energy supplies. Journal of Scientific and Industrial Research. vol 51, pp 257-265,
Mar 1992.
Morthorst, P.E., 1999. “Capacity development and profitability of wind turbines” Energy
Policy 27, pp779-787.
Vector. www.vector.no
SOU 2001: “Handel med elcertifikat - ett nytt satt att framja el fran fornybara
energikallor” Svenska offentliga utredning.
IEA, 2000 : Experience curves for energy technology policy. IEA 2000
Vogstad, K-O, Belsnes, M. M., Tande, J.0.G., Homnes, K.S., Warland, G. (2001):
Integrasjon av vindkraft i det norske kraftsystemet. Sintef TR A5447 EBL-K 32-2001
Vogstad, K-O. (2000a) Utilising the complementary charateristics of wind power and
hydropower through coordinated hydro production scheduling using the EMPS model.
Proceedings, Nordic Wind Power Conference, March 2000, Trondheim, Norway.
Wangensteen, I., Botterud, A., Grinden B., (1999): “Power exchange under various
technical, economical and institutional conditions”, Sintef TR A5015 (Available in
Norwegian, only)
Tande J.0.G., Vogstad, K. (1999) Operational implications of wind power in a hydro
based power system. Proceedings European Wind Energy Conference, 1.-5.3.1999, Nice,
France
Midttun, A., Bakken, B.E. and Wenstgp, F. 1996. Price formation and market stability
under different behavioural assumptions: Theoretical reflections underpinned by compu-
ter simulation of liberal free trade in the Norwegian electricity market.
Appendix
The model is available as a powersim file at http://www.stud.ntnu.no/~klausv/systemdy-
namics/
Back to the Top
CC BY-NC-SA 4.0