Supporting Material is available for this work. For more information, follow the link from
the Table of Contents to "Accessing Supporting Material".
Table of Contents
.
Counterproductive environmental policies: Long term versus
short term substitution effects of gas in a liberalised electricity
market
Klaus V ogstad
Norwegian University of Science and Technology (NTNU)
klausv@ stud.ntnu.no www.stud.ntnu.no/~klausv
Abstract
In Norway, the environmental impact of building gas power in a liberalised market
has been the main controversy for over a decade. proponents of natural gas argue natural
gas substitute more dirty sources of electricity generation in the Nordic market, while op-
ponents argue there is no such guarantee and choose to focus on domestic emissions.
Despite several efforts, energy models have failed in resolving this controversy sat-
isfactory. A survey of previous studies using present energy models (EMPS and NORD-
MOD.-T) for decision support is presented. The models have been re-run and their
sensitivity towards specification assumptions examined.
Second part presents a system dynamics model particularly designed to address the
short- and long run impacts of energy policies. Results show that gas power will substitute
some coal in the short term (as argued by the gas proponent’s), but that the substitution
effect is modest. When including long-term substitution effects of new investments, gas
power also substitute future investments in renewables which results in a net increase in
CO -emissions in the long run. These findings raise serious questions about the environ-
mental benefit of the fuel substitution strategy.
1 Introduction
A remarkable debate has dominated the Norwegian energy policy discourse over the
last decade:
Will new gas power reduce or increase CO»-emissions in the Nordic electricity
market?
proponent’s of gas power argue that natural gas will replace costly and inefficient coal
plants in the Nordic market, while their opponent's claim there is no such guarantee and
that in fact, the introduction of new renewables will suffer from investments in gas. The
controversy already caused the resign of one Government, and continues to hamper con-
structive dialogues among politicians, NGO’s and industry.
Despite several efforts, energy researchers have failed in convincingly resolving this con-
troversy. Though most scientific reports support the conclusion that gas power reduces
CO -emissions, opinions among researchers diverge. There are two plausible explana-
tions for this:
1. The research question is highly sensitive to the assumptions made
2. The models used do not include all the cause-effect relationships considered to be of
importance; therefore their conclusions are not sufficiently persuasive.
In the following, we will examine this controversy in details. Section 3 and 4 of this paper
provides a background for the gas power controversy in Norway. In section 5, a simple
supply curve analysis is provided. Section 6, 7 and 8 deals with the three electricity mar-
ket models EMPS, NordMod-T and Kraftsim. The two first are presently used for decision
support among utilities and regulators, whereas the latter (Kraftsim) is a new system dy-
namics model developed for the Nordic electricity market (Botterud et al 2002; V ogstad
et al. 2002, 2003 and Vogstad, 2004). Previous simulations are examined and re-run with
different specification assumptions. The results support both 1) and 2) for all the three
models, but to various degrees.
The paper ends with a discussion on the different modelling concepts, their strengths
and weaknesses, and to which extent the CO» controversy can be addressed by the various
modelling approaches.
2 The Nordic electricity market
The Nord Pool area is a hydro-thermal system with a yearly average generation of
390 TWh/yr, where 200 TWh comes from hydro, 100, 60 and 10 TWh from nuclear, coal
and natural gas, and 15 and 6 TWh stems from bio and wind respectively. Renewables
play prominent roles in all the Nordic countries’ stated energy plans. The abundance of
these resources played an important role in industrialising the Nordic countries.
In Denmark, wind energy revived during the energy crisis in the 70ies, and is now
the 3rd largest export industry.
Hydropower in Norway gave rise to its energy intensive industry. The paper and pulp
industry in Finland and Sweden makes extensive use of bio resources, residuals and op-
tions for electricity generation. Nuclear power came into use in Sweden and Finland, but
was prevented in Denmark and Norway.
Denmark relies heavily on fossil fuels, but their previous Energy 21 plan (effective
before deregulation) aims at phasing out fossil fuels in order to convert to a renewable
based energy supply within 2050 (Energy 21). Sweden formulated similar targets for a
long-term sustainable energy supply (NUTEK, 1997).
The present situation of the Nordic power supply is summarised in Table 1. Scenarios
for 2010 are based on several reports (in addition to the above mentioned) according to
energy policy goals of each Nordic country.
Table 1 Generation mix in the Nordic countries 1999. The column for 2010 is the future
electricity mix according to political targets.
NOR SWE DEN FIN Total
Supply 1999-2010 «1999 20101999 2010 1999 2010 1999 2010
Hydro [TWh/yr] 115 63 14.5 192.5
Wind P [TWh/yr] - 3 2 4 35 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/yr] 120 123 143 152 34 37 73 85 370 397
In 1991, the Norwegian electricity sector was restructured into an open market. In 1996,
Norway and Sweden formed the first multinational electricity exchange, and the last
member (Jutland, Denmark) joined in 2000. The power balance market, spot market, fu-
ture- and forward market and green certificate market at Nord Pool provide price signals
for utilities and consumers for both short-term and long-term planning. The demand side
participate in all markets, and so far, the market has tuned out to be a liquid, well working
competitive market. Figure 2 shows the historical development of electricity demand,
prices and reservoir levels since 1996. Y early variation of hydro inflow (up to 30%) may
cause large price variations from year to year.
3 The Norwegian CO, controversy
Natural gas for electricity generation is usually considered to be environmentally
beneficial in most other countries, where more dirty sources of generation is substituted.
We will refer to this energy policy as fuel substitution or carbon substitution. In the Nor-
wegian case, the environmental impact of adding gas power is more ambiguous. If we
look at the national level, domestic emissions increase, as the Norwegian supply comprise
100% hydropower. But since Norway is a part of the Nordic electricity market, we must
consider, at least, the impact of the Nordic electricity supply. In a liberalised market, in-
vestment in new capacity will indirectly lead to some substitution of units in the short run,
through changes in the spot price that impact the operation of the marginal units. propo-
nent’s of gas argue that the marginal units in the Nordic market are the old and expensive
coal fired power plants located in Denmark and elsewhere.
Since Norway struck oil in the 70ies, oil and later on gas has been the main export for Nor-
way. It has also been a goal to develop more land-based industry as a spin-off from the
Figure 2 Historical development of consumption, reservoir level and spot price for the
Nord Pool market 1996-2004. (Source: Nord Pool)
Gwh/10
%
1000
NOK/MWh
consumption
water reservoirs |
———= Spotprice (MCP) |
offshore industry, especially domestic utilisation of natural gas.
In the Norwegian white paper (NOU, 1995), it is a goal to increase the domestic use of
natural gas. On this background, several companies looked into the possibility of devel-
oping gas power plants in Norway.
Naturkraft owned by Statoil, Statkraft and Hydro was given the first construction permit
by Ministry of Petroleum and Energy (OED) in June, 1997. Prior to this decision was an
intense debate, and the application process for the emission permit was delayed until after
the Parliament election the same year. The emission permit was granted by The Norwe-
gian Pollution Control Authority (SFT) in 1999, which was litigated by NGO’s until the
final permit was given by Ministry of Environment (MD) in 2001.
March 9th, 2000 the Bondevik Government resigned after losing 81-79 in a Parliament
vote of confidence over denying permit for Norway’ s first gas power plant, being the first
Government resigning from disagreements on the Kyoto protocol and the issue of CO2-
emissions!,
To this date, the permits given for natural gas plants have still not been utilised. Firstly,
strict environmental requirements were imposed by SFT after the permits were given,
which has been delaying the process. Secondly, the electricity market has not made nat-
ural gas profitable yet. Thirdly, infrastructure investments are needed for some of the
projects, and fourth; liberalisation of the European gas market does not give Norwegian
1. CNN news, 09.03.2000
developers significant advantages over European developers for gas power plants.
We will now look into the arguments made on this controversy that has dominated the
Norwegian environmental discourse for over a decade. Energy models have played a cru-
cial role, in trying to resolve this issue. Despite several efforts, energy researchers have
failed in convincingly resolving this controversy, and we hypothesize the reason being
that 1) the research question is highly sensitive to the assumptions made and 2) the models
do not include all the cause-effect relationships believed to be of importance.
3.1 Gas power proponent’ s point of view
The basic argument first put forth by Naturkraft, was that within the Nordic market, build-
ing gas power would substitute coal in other Nordic countries by the operations of the
market. Thus, gas power will in the end reduce Nordic CO-emissions from a regional
perspective. In the processing of the applications, NVE reached the same conclusion.
Their conclusions were based on model simulations using the EMPS model and probably
NORDMOD-T. In the next round of complaints, OED reaffirmed the conclusions, but ad-
mitted there were some uncertainties related to the results.
In the application from Industrikraft Midt-Norge (IMN) of a gas power plant in Skogn,
SINTEF Energy Research analysed the impact on CO,-emissions. The SINTEF study
concluded that CO»-emissions in the Northern European countries (Nordic countries +
Germany) will be reduced as a consequence of building gas power. Their analysis was
based on the EMPS electricity market model.
In October 2000, the new Stoltenberg Government presented their evaluation of the CO,
controversy, changing focus from Nordic countries a European level. The Government
concluded that CO2-emission reductions were the most likely outcome from building gas
power plants, while this view was contested by the opposition. In addition, the authors
that had provided analyses, criticised the Government for misinterpreting their material!
3.2 Opponent’s point of view
While proponent’ s argue gas power will substitute coal, opponents argue there is no such
guarantee, and that gas power will come in addition to coal power. Opponents also seem
to focus on national emissions and international obligations. They argue that gas power
will increase demand, and that coal power plants elsewhere is not likely to shut down their
plants as a result of the introduction of gas in Norway. They emphasize statements from
SFT?, where itis said that gas power also will delay the necessary transition to renewables
such as bio and wind power.
During the new Governments presentation of the issue in October 2000, an IEA report
showed that development of new gas plants will continue to grow in EU, without replac-
ing existing coal plants. The EU minister of Environment, Domingo Jimenez-Beltran,
rejected the Norwegian Minister of Environment's statement? that claimed Norwegian
gas power substitute European coal power. No models were involved in the NGO’s anal-
1. Interview with T. Bye (Statistics Norway) in Dagbladet, 31.10.2000
2. National Pollution Authority
yses.
From the above discussion, it appears that the proponent’s focus on short-term ef-
fects, such as short term substitution coordinated by the operations of the market. Com-
parative static economics and detailed production scheduling models such as the EMPS
model provide tools for analysing these interrelationships. The opponents however, seem
to focus on the longer term aspects, and tend to ignore the short-term effects. They con-
sider replacements of investments when speaking of new developments, and even in the
longer term about technology progress. There were no model studies however, that incor-
porated these effects.
None of the groups seem to consider both the short term and the long term aspects (i.e.
both substitution effects of generation scheduling, substitutions in investments decisions
and so forth). Furthermore, geographical system boundaries are inconsistent in the dis-
cussions and in between the model studies. Opponents focus on national emissions, while
proponents usually consider the Nordic countries plus power exchange with Germany.
4 A simple analysis of supply curve and market prices
In the Nordic market, electricity generation is scheduled in the short term by short
run marginal costs. This information is not readily available in a competitive market, so
any information on costs is guesstimates afflicted with uncertainties. Figure 3 shows the
supply curve of the Nordic electricity market that has been used in our EMPS simulation
runs and earlier versions of the Kraftsim model (V ogstad et al., 2002). Hydropower, wind
power and exchange are not included in the supply curve. Nord Pool’s spot price distri-
bution for 2001 is shown in the same graph. Held together with the supply curve, the data
shows a picture that does not quite match the assumptions of coal being the only genera-
tion technology replaced by gas. From the supply curve, coal serves as baseload well be-
low the average spot price level. Among baseload units are also CHP (including bio),
nuclear and natural gas units operating at marginal costs below spot prices. In the range
of the spot price distribution, we find some coal, oil, bio and gas. Peak load gas turbines
and backup-coal can be found well above the price distribution range, suggesting that the
inefficient and costly coal fired units are not frequently in use. The picture is thus more
complex than assuming coal to be marginal generation. Rather, inspection of the graph
and the production data (see A ppendix 1) indicates that new gas power replaces existing
gas power (as well as coal and oil) in the Nordic market.
This supply curve analysis does however not provide the complete picture. Firstly,
exchange is not accounted for, and capacity constraints for transmission between coun-
tries are not included. Furthermore, hydropower with reservoirs is not adequately repre-
sented in a supply curve as the water values change with changes in reservoir level
content. On a yearly basis however, hydro schedulers try to schedule generation in order
to maximise profits while avoiding spillage. To include such considerations, electricity
market models have been developed that simulate the behaviour of the market. These
models have also been used to address the CO, controversy. In the following we will ex-
amine simulations analyses by the EMPS model and NORDMOD.T. The new system dy-
namic model Kraftsim, is meant as a complement to existing decision support tools, both
3. Interview with Domingo Jimenez-Beltran, (EU Minister of Environment) in Dagbladet
25.10.2000
Figure 3 Supply curve, emission intensity and spot price distribution in the Nordic
electricity market. The spot price distribution was calculated from hourly time series for
the Nord Pool market in 2001.
a)
1 . 1400
09 { | Back-up coal
OF ' 4
--------- 4 -- gee i iT. - -- Gas Turbines peak load _____]4200
08; Natural obs |
S07 17 = possssa aa nan 41000
S Baseload coat =
le fe) Ce ae ba-------- es 4800
ee 1 1 >
—€05 ! ' 4 2
eee ee eee Ah soagessGemsse=ess loesemesen 6005
= aI SEY ig
§ 0.4 ; ; 3
Gos, eed af —-i}—--— La-------+ —_—— 4400 §
\ ' oO
02 ' ' Tene
a + ~~ 7F distritoption \- ~~~ ~~~ ~~~ p=} === 200
0.1 Hl 2001 | Hl — capacity
i t t « |4 emission
0 L i}
0 100 300 400 500
cost [NOK/MWh]
for utilities and regulators. Table 4 summarise the three model characteristics and their
differences. In the subsequent sections 6 to 8, we will examine the simulation runs that
address the CO emission controversy.
Table4 Overview of model characteristics
Model EMPS NordMod-T Kraftsim
Purpose Optimal hydro sched- Policy analysis, max- Policy analysis
uling and price prog- imises socio-eco-
nosis nomic surplus
Type Technical bottom-up, Technical bottom-up, System dynamic with
partial equilibrium. partial equilibrium. _ focus on competition
Stochastic dynamic Optimisation of between energy tech-
optimisation of socio-economic sur- nologies
hydropower genera- plus
tion
Time horizon 1 year <20 yr <30 yr
Spatial resolution 12 areas (Nordic 4 areas (Nordic coun- One area (Nord Pool)
countries+Germany) tries)
Electricity price Endogenous Endogenous Endogenous
Demand! Endogenous Endogenous Endogenous
Generation scheduling Endogenous Endogenous Endogenous
Capacity acquisition Exogenous Endogenous Endogenous
Resource availability Exogenous Endogenous for Endogenous for
hydropower renewables
Technology progress Exogenous Exogenous Endogenous for
renewables
1. Demand growth rate is exogenous, while price elasticity of demand is endogenous
5 Analysing CO,-emissions with the EMPS model
EMPS (Efi’s Multi-area Power Simulator) is a decision support tool for seasonal hydro
scheduling. Though it was originally developed for hydro scheduling purposes and price
prognosis (Fosso et al. 1999), itis also used for energy policy studies
The model is a technical bottom-up model containing a detailed representation of the hy-
draulic system of reservoirs and generating units. The supply side is described with indi-
vidual plants within each area. The stochastic representation of hydro inflow utilise 60-
70 years of historical inflow data. The model optimises hydro generation over a year us-
ing stochastic dynamic programming and the water value method. Main features and ex-
ogenous versus endogenous variables are displayed in Table 4. Electricity price and
generation scheduling is endogenous, while long term mechanisms such as capacity ac-
quisition, technology progress and resource availability does not need to be represented
within the one-yeartime horizon. Figure 5 shows an overview of the physical description
of supply and demand within each area. The graphs show the optimal reservoir level
curves, and the resulting prices. The results are shown as percentiles emphasizing the sto-
chastic optimisation of hydro scheduling with stochastic inflow.
The EMPS model has been used to analyse the impact on Nordic CO -emissions
from building new gas power plants (Wangensteen et al., 1999) Sintef Energy research
provided the impact study of changes in Northern-European CO»-emissions from build-
Figure5 The EMPS model consists of several interconnected local areas with various
supply technologies, demand and market access. (Source: Vogstad et al, 2001; Vogstad
2000)
MidtNorge area
ing 800 MW gas power in Skogn papermill, located 100 km’s north of Trondheim.
The results are reported in W angensteen et. al (2000) and in the consequence report! Fig-
ure 6 shows a CLD representation of the EMPS model. As can be seen, Capacity is ex-
ogenous to the model. Consequently, investment substitutions must be handled
exogenously. The power exchange loop (B3) represent exchange between areas. The ex-
change depends on the available transmission capacity between the areas, and the price
difference. The market clears generation and demand for each time step”. Thermal gen-
eration is based on marginal costs (MCjy), whereas hydropower and wind power differ in
this respect. Wind generation is stochastic (represented by 30 years of historical data),
and hydro inflow utilise 60 years of historical data in its stochastic representation. Hydro
generation is based on the water value principle, in which a value of storing one additional
unit of water is derived from a stochastic dynamic optimisation of the expected future
1. Available online www.industrikraft.no
2. Time resolution is one week, but demand can he subdivided into load blocks (usually 4) for
within each week.
10
Figure 6 CLD representation of the EMPS model
Stochastic wind j
Stochastic inlow j Xt
wind generation j -
ed, Transmission
eS loss jk Transmission
Reservoir con capacity i
hyd tio c
BA- Reservoir oe) emission fj
~ drawdown + | y hey A
water value j SO them total generation j xchange.)
f_ /*, generation j B3 - Power
Capacty j Exchange
B1- Generation Je price
capacity factor ij Scheduling - yy + difference jk
aif & Price j
BO-
MC j *Demand -
B2- Sh
ean Demand j
Scheduling
i- technology of type i
j,k - region index
profits over the time horizon (Vogstad, 2004). The interdependency of hydro generation,
reservoirs and spot price is illustrated by the Long term scheduling and the Reservoir
drawdown loop.
Table 7 shows the concluding result from the Skogn analysis by SINTEF Energy Re-
search using the EMPS model It was concluded that adding 800 MW gas power in Skogn
would increase domestic CO. - emissions by 1.9 Mt/yr, while emission reductions take
place in other Nordic countries and in particular Germany. The result is a net reduction
of 1.1 Mt CO» peryear. As can be seen from the tabulated values, differences are small
in comparison to the total emission values, which suggest the analysis to be highly sensi-
11
tive to assumptions made.
Table7 —_ Results from the Skogn study using EMPS (Source: Sintef Energy Research, 2000)
All numbers in Mt CO,/yr Without gas power With gas power plant Difference
plant
Norway 21 4.0 +1.9
Denmark 23.3 22.9 -0.4
Sweden 8.8 7.9 -0.9
Finland 40.8 40.5 -0.3
Germany 366.3 364.9 -14
SUM Nordic+Germany 441.3 440.2 “LT
Table8 EEPS simulations re-run with various data sets and assumptions change in CO -
emissions
Scenario Nor Den Swe Fin Ger Tot
Skogn2005 18 -0.3 -0.9 -0.3 -14 “Ll
1999 23 “11 -0.6 -1.0 -1.8 2.2
ref2010 3.2 -0.1 0 13 -2.9 “11
wind2010 24 -0.3 -0.1 “13 -24 “17
noexchange2010 2.2 -0.6 -0.4 -2.0 0 -0.8
newdata2010 203 -0.7 -1.3 -1.2 13 -2.0
noboilers2010 2.5 -0.7 0 “11 -1.3 -0.6
In Table 8, new simulation runs have been performed to assess the robustness of the re-
sults compared to the Skogn study. The scenarios are as follows:
Skogn 2005 - This scenario is taken from the Skogn study (Sintef report), where there is
a weak growth in demand (1.2%/yr) towards 2005 and some new transmission capacity
(600 MW) to Germany is added.
ref1999 - Nordic situation as of 1999, with the data set in shown in Figure 3 correspond-
ing to the installed capacity in 1999. The resulting CO7-emissions from this scenario cor-
respond well with actual CO>-emissions for that year (V ogstad, 2000). (See A ppendix 1)
ref2010 - Scenario 2010 without new wind power, as defined in Table 1
wind2010- With 16 TWh/yrwind power according to each country’s plans. (see Table 1)
noexchange2010- Scenario as for wind 2010, but without exchange to Germany.
newdata2010 - Scenario with new data set for Germany based on Bower et al (2000)
noboilers2010 - Same as newdata2010, but substitution reduction on demand side (i.e.
electrical boilers) omitted.
The scenarios ref1999 and wind2010 scenarios are also documented in Vogstad et al.
(2000).
We will shortly comment upon the above tabulated results. The results clearly show the
short-run substitution effect for all of the scenarios. The major share of substitution takes
place in Germany, followed by Finland. Some of the results will be commented upon in
the following. A large substitution effect is seen in 1999 compared to the scenarios for
2010. Especially in Denmark, fuel switching from coal to gas is scheduled, as new coal
12
poweris prohibited, which results in lower substitution effects of CO. in the 2010 scenar-
ios. A larger share of the substitution is then moved to Germany. The difference between
ref2010 and wind2010, is the addition of wind from 4.5 to 16 TWh according to the Nor-
dic countries wind energy goals in 2010. The increase in substitution effect between these
scenarios is due to substitution on the demand side. In the noexchange2010 scenario, we
only removed the possibility for exchange to Germany, which results in increased substi-
tution within the Nordic countries. The result shows a significant reduction in Finland,
due to some of the Finnish coal plants. In newdata2010, a new data set for Germany is
used, based on Boweret. al (2000). The results yielded more CO, - reductions in Sweden
due to more imports from Germany. The last scenario, noboiler2010 shows the same re-
sults when the substitution effects from oil/el boilers and other demand side flexible loads
are not accounted for. This sensitivity analysis shows that the main substitution effect is
actually on the demand side, where cheaper electricity prices result in fuel switching from
oil to el in flexible boilers. The uncertainty of the installed oil/el boilers and their opera-
tions (depending on changes in oil taxes etc.) is considered to be substantial.
However, all the scenarios show reductions of CO, from building gas power in Nor-
way. Most substitution takes place in Germany, thereafter Finland, while the substitution
effect in Denmark and Sweden is less significant.
Two data sets for Germany were tested, and the latter is believed to be more updated.
Based on demand and supply provided by the data set, however, electricity prices in Ger-
many should be around 90-130 NOK/MWh, as calculated by the EMPS model. The ob-
served prices in the European Energy Exchange! (EEX), are however much higher (170
NOK/MWh in 2000, and 240 NOK/MWh in 2003) without any significant changes in the
supply ordemand. An explanation for these high prices is provided in Bower et. al (2000)
as strategic bidding enabled by increasing market concentration. Observed market prices
and data on supply/demand and marginal costs of generation does therefore not match,
which poses a dilemma for all of the three models if we are to assess the environmental
impact of import/export to Germany.
The benefit of using the EMPS model, is the good description of hydro scheduling
and price formation in the Nordic market. The disadvantage is that the long-term effects
such as investment substitutions of capacity acquisition is not included in the model and
must be assumed for each scenario.
6 CO,-emission analysis using NORDMOD-T
Both generation scheduling and investment decisions are endogenous in NORD-
MOD-T, and analyses using this model should therefore also include effects of investment
substitution. Figure 9 shows the generation scheduling, power exchange, capacity acqui-
sition and resource availability feedback loops. Investments in a technology are made if
long-run marginal costs are lower than the market price for the next time period. Capacity
is then added the next period (investments are made at the start of each year). There is
also a maximum constraint on the amount of capacity from each technology that can be
added.
The model is also a detailed bottom-up description of technologies, using load dura-
tion curves and blocks that characterise four load modes for four seasons. Aune et al.
(2000) summarise their findings in their studies. Some aggregated results are shown be-
1. For price information at European Energy Exchange see www.EEX.de
13
Figure9 CLD representation of NordMod-T
Transmission
loss jk Transmission
C02 emission j capacity jk
7 + th
— total generation j exchange jc
Capacity i generation jj
= B2 - Power
J B1- Generation Exchange price
BS. capacily factorij Scheduling - 4 + aiteenre je
Resource . 4
availability MC j—*- J
Profiabity Ba. Capacity Demand j
acquisition
LRMC §
i- technology of type i
j,k region index
low:
The study analysed high, low and medium price scenarios for Europe, while coal was
assumed to be the marginal unit of generation in Europe.
However, if prices are high, gas power is more likely to be the marginal unit in Europe. It
tured out that investments in wind power was exogenously determined, so eventual sub-
stitution effects of renewables only consider biomass.
Assumptions of transmission capacity and non-Nordic electricity prices are shown in
Figure 10 Left: Assumptions on tranmission capacity to non-nordic countries. Right:
Price scenarious for non-Nordic countries, base run. (Source : Aune et al. 2000)
1600 3 ]
‘400 x
1200 D re /
1000 Nor | 25 [Bese
800 —SWE | 20 —Medium|
ca | 8 eS
400 10
200 5
o °
FLIES L ESS PEL S ELSES SP
Figure 10 for the NordMod-T simulations. Figure 11 shows the development of CO, -
emission from adding 5.6 TWh Norwegian gas power in 2004 for various assumptions
of non-Nordic electricity price; Low, medium and High prices. Low prices are 80, 110
and 140 NOK/MWh for base, medium and high block; Medium price scenario is 100
NOK/MWh for baseload, and correspondingly +25% and +50% higher prices formedium
and high block. The high prices scenario assume 150, 188 and 225 NOK/MWh for base
14
block, medium and high block prices.
Figure 11 Changes in CO, - emissions from adding 5.6 TWh gas power in Norway in
2004. (emission changes in non-nordic countries included). The three scenarios include
Low, Medium and High non-nordic electricity prices. (Adapted from Aune et al., 2000)
Change in CO2-emissions
2 —< Low price
— Medium price
-@- High price
Mt CO2/yr
2
2004 =2005 2006 2007 2008 2009 2010
year
The study concluded that there is high uncertainty whether building gas power in
Norway increase or reduce Norther European CO-emissions, and that the results rely
heavily on the assumptions made, in particular the price level in Europe, and the available
transmission capacity to Europe. If transmission lines were congested so that Norwegian
gas power would substitute generation in other Nordic countries, gas would substitute gas
and hence there could even be increased CO» emissions.
7 CO>-emission analysis using Kraftsim
The Kraftsim model was developed to analyse long-term versus short-term consequences
of energy policies within the context of a liberalised Nordic electricity market (V ogstad,
2003; 2004). The time horizon is 30 years, and the time resolution sufficiently captures
features of generation scheduling at a seasonal and weekly level!. The Nordic market is
represented as one area, and the model has no spatial disaggregation. The model focuses
1. The smallest time constant is 3 days for spot price adjustments, in order to clear supply and
demand with a weekly load variation. The numerical time step is 1 day. To capture daily load
pattern, spot price adjustment time and the numerical time step can be adjusted down to an
hourly resolution. This will be done when the effect of start/stop costs and ramp-up constraints
are included for each generation technology (i.e. the unit commitment problem)
15
on the competition between the following main technologies i:
nu - nuclear
co - coal
ga - natural gas
gc - natural gas with CO, sequestration
gp - natural gas peak load;
hy - hydro
bi- bio
wi - wind onshore
wo -wind offshore.
The main loops of Kraftsim is shown in Figure 12
Figure 12 Kraftsim CLD diagram
Fractional growth
‘80 - Demand
balance
Re Price elasticity of
Price of electricity. demand
B1- Generation
Electricity scheduling
generation iv 4 Capacity factor iv
a
+
CO2 emission iv _ B2 - Capacity
R acquisition
_——™ operational
Capseiyiv- + iiency iv costs iv
- (3 vintages) +
Resource F Fuel costs
availability i
+ new capacity i +
ee + Expected profitability
iti -Leaming of new capacity i
B3 - Resource Surve
depletion R i- technology of type i
Investment and
operational costs i v - vintage of type v
B1 - Generation scheduling. On a daily basis, electricity generation is scheduled by mar-
ginal costs of operation. The last unit in operation determine the spot price at each time
point (in a uniform-price auction, perfect market). In this model, the supply is described
by each of the nine technologies i, their vintage v and fuel costs.
B2 - Capacity acquisition is the process of investing in new capacity based on the expect-
ed profitability of new capacity. Expectations of future electricity prices play a crucial
role in this case. If the expected future electricity price sustains at levels higher than the
16
long run marginal cost of new generation, new capacity is added.
R1 - The learning curve effect is a reinforcing loop. As more capacity is developed, the
technology and know-how progresses, reduces the costs and increase the profitability of
new capacity.
B3 - Resource depletion finally constrain expansion of new capacity. All resources are
constrained in terms of available land, riverfalls or fossil reserves. As more resources are
utilised, costs of utilising the remaining resources increase.
All decisions governing the operations and investments in technologies occur in a com-
petitive market. Short term prices govern generation scheduling (B1), investment deci-
sions are based on profitability assessments (B2) and resources and technology progress
(R1) is partly endogenous to the model (compare with Table 4).
This paper reports of a specific policy study using the Kraftsim model and we will
only briefly present the most important assumptions underlying the model. A complete
documentation of the Kraftsim model can be found in Vogstad (2004)!.
All the decisions are made in a competitive environment.
7.1. Generation scheduling
Other electricity markets such as the German, Dutch, Spanish, UK, and Californian
market are characterised by some few, dominating market players. In contrast, the
number of market participants in the Nord Pool market is fairly large, and regarded as
highly competitive’. Itis therefore assumed that market participants bid into the spot mar-
ket according to their marginal costs (i.e. a perfect spot market). This assumption is in
accordance with the two previously mentioned models.
Generation scheduling
(1) CF,,= fj,(Price/operational cost) (1]
(2) operational cost;, = Fuel cost; /resource efficiency; [NOK/MWh]
where fiy(.) is a table look up function that has the shape of a cumulative density function.
The sum of all technologies i for all vintages v then represent the aggregated supply curve
for thermal technologies. The marginal costs of hydropower are calculated by the water
value, while wind generation is determined by the wind conditions.
7.2 Investment decisions
Investments are purely based on a Retum on Investment criteria (ROI) for profitability
considerations using net present value calculations. The required return on investment
uses an interest rate of 7%, which is the recommended interest rate for socio-economic
calculations. In a competitive environment, utilities require higher interest rates and
1. Available at www.stud.ntnu.no/~klausv under publications (forthcoming)
2. Hansen et al. (2001) argue that historical observations of the Nord Pool market may be mislead-
ing in the evaluation of market power. Nord pool inherited a power system with excess capacity
from the regulatory regime. There is a trend in mergers and acquisitions. With increasing mar-
ket concentration, market power may become a problem in the near future
17
shorter pay back periods in their profitability assessment. The resulting investments are
therefore considered to be in the optimistic range.
Price expectations play a crucial role in the profitability assessment of a generation
technology. The futures market at Nord Pool represents the best available information on
the joint expectation of future electricity prices up to 4 years ahead. Y et, investors need
to consider longer time horizons than just 4 years ahead and need to take other information
into account. Investors can then look at the long-term fundamentals of the supply and de-
mand. A convenient rule is to assume that the electricity market will converge towards
long-term equilibrium at which the long run marginal costs of the least expensive technol-
ogy sets the market price!. But this type of information is also uncertain, as it for instance
relies on fuel price expectations.
On the other hand, future markets are influenced by conditions of the present, such as two
consecutive dry years resulting in low reservoir levels, cold winters, or similar occurrenc-
es that will even out in the long run. We therefore assume the investor to pay some atten-
tion to the futures market, and some attention to the long-run marginal costs of new
generation as described in Eq (3) and (4) below:
Profitability assessment
(3) | Expected future price = Weight on LRMC -min,{LRMC;} + (1-Weight on LRMC)-Futures
price [NOK/MWh]
(4) Weight on LRMC = 0.6 (1]
The effect of profitability on investment rate multiplier governs applications and invest-
ment decisions, based on the a (return on investments to required return on invest-
1. Statements by executives and interviews in media suggest that investors use this rule when
looking beyond the futures market.
18
ments ratio):
(5) effect of profitability on investment rate; = f(ROI,/RROI;) [1]
(6) RROI,= Lifetime; ‘annuity factor; [1]
(7) annuity factor; = Internal rate of return/(1-(1+Internal rate of return) Lifetime;) (]
(8) Internal rate of return = 7 [%/yr]
(9) ROI; = (Expected future price - operational costs, + O&M; - Incentives,)/Energy invest-
ment costs; [1]
(10) Energy investment costs, = Investment costs; / (Expected CFj-Full load hrs; ‘Lifetime; )
[NOK/MWh]
(11) Investment costs; = Initial investment costs; - learning multiplier, [NOK/kW]
(12) operational costs; = Fuel cost; / Resource efficiency; [NOK/
MWh]
(13) Incentives = {0,0,0,0,0,0,100,100,100} [NOK/MWh]
(14) Lifetime; = {40,30,30,30,30,40,30,20,20} [yr]
(15) CF estimated; =f,(Price/operational costs;) [1]
(16) Yearly average CF; = SLIDINGAVERAGE(CF estimated, , 1 yr) [1]
(17) Expected CF, = DELAYINF(CF estimated, 3 yr) [1]
( ]
18) Fuel costs; = {26.4, 47,80,80,80,0,80-effect of resource on fuel costs bi,0,0} [NOK/MWh:
Where f(.) denotes a table look up function, and Full load hrs; , Resource efficiency; and
learning multiplier; is defined elsewhere in the model (see Figure 14). Figure 16 shows
the development of LRMC for each technology that is endogenously computed by the
model. (The initially high LRMC values for gas with CO sequestration (4) and gas peak
load (5), is the very low expected capacity utilisation of these technologies at low elec-
tricity prices, see Eq (15)).
7.3. Technology progress
Technology progress is difficult to endogenize in a regional model, since much of the
technology progress usually occurs at a global level.
However, Danish wind turbine manufacturers are among the world leaders. The ear-
ly stages of wind turbine development can largely be attributed to the development in
Denmark, and are now taking the lead in developing offshore wind parks in the shallow
waters surrounding Denmark. The Nordic countries all have good resources for further
wind power development.
Sweden, Finland and Denmark all have a strong foothold in bio energy. A large pa-
per and pulp industry has provided favourable conditions for bio energy to develop in both
Finland and Sweden, whereas residuals from the large farming industry has motivated
RD&D! of bio energy in Denmark.
Norway has strong traditions in hydropower technology. Hydropower is however a
mature technology and there is less potential for improvements, but there are still ad-
vancement in the development of small scale hydropower. Local adaptations have to be
1. Research, Development and Deployment
19
done for bio energy concerning resource base, infrastructure and industry.
We could therefore justify leaming to be endogenous for the renewable technologies, al-
though learning can also be represented exogenously.
In the case of thermal generation technologies, the learning effect is taken as exoge-
nous as the major environments and markets for thermal generation technologies are out-
side the Nordic countries.
7.4 Resource availability
Prices on nuclear, coal and natural gas are assumed to be fixed during the simulation
period. This assumption is rather conservative with respect to the price of fossil fuels.
Most scenarios for fossil fuels indicate rising prices, in particular for natural gas. The as-
sumption of natural gas prices in the Nordic countries being independent on the construc-
tion of gas power could also be questioned, so the development of gas power is rather
optimistic in our model.
There is a feedback from hydropower, bio and wind resources to the costs of developing
new resources. For each project developed, less attractive sites must be utilised. An ex-
emption is offshore wind power, for which we assume there to be neglible feedback to
costs during the time period considered in our model.
7.5 Demand side
Demand side is kept simple in this model. We account for an underlying growth trend of
1.5%/yr, a weekly and seasonal variation’. In addition there is a price elasticity of de-
mand (0.3 1/yr) that reflects improvements in energy efficiency or new investments on the
1. Actually, daily load variation is more important than the weekly variation, while seasonal varia-
tion is the most important. The model can increase resolution to capture hourly variation, but
will involve more model development on the supply side.
20
demand side.
Figure 13 Electricity demand profile organised into seasonal and hourly variation
(Adapted from Nord Pool, 2001).
7.6 Capacity acquisition and vintage structure
On the basis of profitability assessments, investors submit applications to the authorities.
The application processing takes time, depending on the technology. The final invest-
ment decision is made later on, after permits have been obtained. The application process
takes from one to several years, and construction involve significant time delays as well.
Capacity has been divided into three vintages v: new, intermediate and old. Each vin-
tage is characterised by its resource efficiency. Old coal plants are typically less efficien-
cy than new ones. The continuous replacement of old plants with new, more modem
plants increase efficiency of the capacity stock, and consequently the supply curve of gen-
erating units and related CO»-emissions.
The corresponding stock and flow diagram is shown on next page
21
Figure 14 Kraftsim model SFD diagram
22
7.7 Simulation results
To test the system response of the fuel substitution strategy, we introduce 3200 MW of
new natural gas in 2005. This simulation run is compared to a reference run in the fol-
lowing graphs. The reference run displays the evolution of the Nordic electricity market
towards 2030 in terms of electricity price development, investments, generation mix and
finally CO2-emissions. In all simulations, a subsidy of 100 NOK/MWh is provided to all
renewables technologies except hydropower. The resulting data are smoothed to yearly
averages, while the underlying simulations include seasonal variations.
Figure 15 Spot price development for the reference case (*) and the fuel substitution
scenario introducing 3200 MW natural gas in 2005.
NOK/MWh.
Overcapacity reduction
500 —
400
“FE price
—*Price
2 yearly average price
3 *yearly average price
200
A+3200 MW natural gas in 2005
o+ t + + + + 1
01 Jan 2000 01 Jan 2005 01 Jan 2010 01 Jan 2015 01 Jan 2020 01 Jan 2025 01 Jan 2030
7.7.1 Electricity price development
The observed development in the reference run deserves some explanation. In Figure 15
the spot price (1) is shown. The rapid fluctuations (1) are caused by the seasonal and
weekly variations in demand, which is quite significant in the Nordic market due to a sub-
stantial share of electrical heating and the seasonal inflow of hydro. To easier identify
price trends, the yearly average price (3) is plotted as a sliding yearly average. In the ref-
erence scenario, we observe an increasing price towards 2015, whereas prices show a de-
clining trend towards the end of the simulation period. Towards the end of the simulation
period, prices exhibit long-term oscillations.
The increasing price trend towards 2015 is due to the initial overcapacity in the Nor-
dic market. The capacity acquisition loop drives the market towards long-run equilibri-
um, so that the long-run electricity market prices approach the long-run marginal costs of
new generation. If we compare the futures price with the long-run marginal costs
(LRMC) of new generation in Figure 16, we see that the futures price will converge to-
wards LRMC for gas power and, in the long run, offshore wind power. The market price
converges to LRMC for the cheapest technology on LRMC and futures prices (see chapter
7.2) - depending on investors’ weight on LRMC and futures prices. For more details on
the price development, see Notes a the end of the paper.
The price response to introducing 3200 MW natural gas in 2005 is shown as the bold
line (2) in Figure 15. Obviously, the introduction of new gas power suppresses electricity
23
Figure 16 Future prices versus long run marginal costs of generation technologies
NOK/MWh
oot
“4 LRMCInu]
6007 > LRMC{co}
3 LRMCigal
5004+ + LRMCigc]
a 5 LRMCIgp]
aot 5 — LRMCIhy]
& LRMC[bi}
08 6—F 2 LRMC{ wi)
a ———— a 8 LRMC[wo]
tot > Futures price
Pe 8 9 Pi
200 ae SS A 6 LRMC with incentives{ bi}
h— eg 2 LAMC with incentives{ i]
100+ TO 8. LRMC with incentives{wo]
° + + + t + 1
01 Jan 2000 01 Jan 2010 01 Jan 2020 01 Jan 2030
prices. Introducing 3200 MW in a system of 80 000 MW also triggers long-term price
oscillations, which in tum can cause boom/bust cycles in the acquisition of new capacity.
Although an interesting result itself, oscillations are not the focus of this study. (See
Notes for extended discussion).
7.7.2. Substitution effects in capacity and generation
Figure 17 shows the development of capacity for the reference run (thin lines) and the fuel
substitution scenario (bold lines). The reference run shows a steady growth in natural gas
and wind power. At the end of the time period, offshore wind power becomes significant,
while bio energy does not show significant growth. The hydropower resources are al-
ready fully utilised, whereas nuclear and coal is phased out due to their low profitability!.
Peak load capacity is also being phased out, as it is not profitable to invest in peak load
capacity purely from electricity price considerations.
The bold lines shows the fuel substitution scenario, where 3200 MW natural gas is added
in 2005. The immediate system response in capacity development does not differ signif-
icantly from the reference run, but as the simulation progresses, new investments in bio,
wind and offshore wind are systematically reduced compared to the reference run. Thus,
investments in gas substitute new investments in renewables in the long run.
If we now consider generation scheduling, Figure 18 shows the (averaged) yearly
generation for each technology. As can be seen, coal (2) responds slightly by reducing
its capacity utilisation when 3200 MW natural gas is added in 2005. The marginal costs
of coal are, however well below the new market price trajectory, and the substitution ef-
1. Uncertainties of COy-quota prices make coal less attractive as well. In Denmark, new coal
plants cannot obtain construction permits. Sweden decided in 1980 to phase out their existing
nuclear capacity, but so far only 600 MW of the capacity has been phased out. On the contrary,
Finland recently decided to expand one of their nuclear plants. According to NVE, investment
cost for the new Finnish plant was reported to be 13 KNOK/kW (NVE 2002 p22), while average
investment costs of nuclear plants are 22.5 KNOK/kW in the same report. The increased focus
on risk in a competitive environment also make these investment-intensive technologies with
long lead time less attractive.
24
Figure 17 Capacity development. The investment substitution effect of adding gas
power
mw
40,000+
30,000-+ gc
20,000:
total capacity
10,000
of a
O1jan 2000 O1jan2005 O1jan2010 O1jan2015 O1jan 2020 O1jan 2025 01 Jan 2030
fect from coal is therefore modest. Exports increase, which substitute coal abroad as well.
The marginal costs of coal are typically in the range of 100 NOK/MWh before the capac-
ity utilisation of coal is significantly reduced. Hydropower also responds to the added ca-
pacity of gas. In hydropower generation, the water values! are compared to the spot price.
If water values are lower than the current spot price, it is more profitable to release water
than store the water for later generation. Water values are however, regularly being up-
dated when new information arrives on inflow, consumption or new capacity. It takes
some time before all the utilities involved in hydropower generation incorporate new in-
formation into their production planning tools (such as the EMPS model). Reservoir lev-
els can, in addition to seasonal variation of inflow, absorb variations in generation from
year to year, but usually not more than three years.
The reduced generation corresponding to reduced investments can be observed for
bio, wind and offshore wind (see bold line 7,8 and 9) in Figure 18.
7.7.3. Long run versus short run effect of the fuel substitution strategy on CO>-
emissions
With respect to CO»-emissions, the consequence of introducing gas power has both short
run and long run implications. In the short run, CO» emissions from coal and peak load
turbines are reduced, but this effect is modest as discussed in the previous section. The
increase in exports (negative values) compared to the reference run significantly contrib-
utes to reduce CO,-emissions. This contribution is also accounted for in the total emis-
sion rate, and as argued by proponent’s of gas power, we can observe a short-term total
CO>-reduction.
Thus, gas power substitute generation some generation from coal in the short run. Asa
very conservative assumption, we assumed the marginal electricity generation from the
continent (Germany, Poland and the Netherlands) to be coal with the least efficient tech-
1. Water values reflect the marginal value of storing one additional unit of water
25
Figure 18 Yearly generation. Short run substitution effects in generation of adding gas
power.
Twhiyr iglyauriy aug generaGontnul
200 ee ng GR 4-+yearly avg generation{ nu]
2 yearly avg genétationfes)
2 * yearly avg generation{co]
= yearly avg generation{ ga]
asot |3- *yearly avg generationigal
4- yearly avg generationigcl
4 *yearly avg generation{gc]
5- yearly avg generation[ gp!
5-*yearly avg generation{gp]
-& yearly avg generation{hy]
-& *yearly avg generation{hy]
2- yearly avg generation{bi]
7 *yearly avg generation{bi]
2. yearly avg generation{ wi]
& *yearly avg generation{wi]
}9- ye Orly avg generation[ wo]
-9- *yeary avg generation{wo]
‘ ; ; 40 yearly avg exchange
01 Jan 2020 01 Jan 2030 [40 *yearly avg exchange
50
0.
01 Jan 2000 01 Jan 2010
nology. This conservative assumption provide an upper bound scenario for emissions ac-
companied by imports, but even in this case - total CO2 emissions increase in the long
term! The substitution effect of gas towards reducing coal in the Nordic countries and
through exchange does not compensate for the long run substitution impacts on invest-
26
ment in renewables and the long term stimulation of demand increase.
Figure 19 Change in CO-emissions from building gas power compared to reference
run (*)
Mtonne/yr Short term substitution >08g term substitution
of coal of investments
1 in renewables,
-+ total yearly CO2 emission rate
4.*total yearly CO2 emission rate
“2 yearly CO2 emission ratelco]
-2 * yearly CO2 emission rate{co]
3 yearly CO2 emission rate[gal
3 * yearly CO2 emission rate[ ga]
~4 yearly CO2 emission rate[gc]
-4 * yearly CO2 emission rate{gcl
-5- yearly CO2 emission rate(gp]
-5 yearly CO2 emission rate{ gp]
6 yearly emission from exchange
~6-* yearly emission from exchange
01 Jan 2000 01 Jan 2010 01 Jan 2020 01 Jan 2030
8 Structural- and parameter sensitivity of the simulation results
8.1 Parameter sensitivity
Various scenarios were tested for the EMPS model simulation that gave different lev-
els of CO2-emission reduction, but each result gave a net CO>-reduction.
The NordMod-T study contained several scenarios with low, intermediate and high
relative prices between EU and Nord Pool. The results showed that 1) the Transmission
capacity was important for the result, and 2) that there was no certain impact of CO»-emis-
sion from adding gas power in Norway. The study emphasised the significant uncertainty
related to the results.
In the Kraftsim case, some additional simulation runs were performed to assess the
robustness of the results. Assumptions were also made conservative, i.e. it was assumed
that exchange to the continent would replace old coal fired units. Another extreme sen-
sitivity test was to rule out technology progress as uncertainties of the learning curve ef-
fect could yield too optimistic results on development of renewables. However, the
results still showed significant increases in CO-emissions when adding gas power.
8.2 Representing transmission constraints
One of the main differences between the three models, are the spatial degree of spatial dis-
aggregation. The EMPS model is the most detailed in this respect (12 regions) while Nor-
dMod-T divided the Nord Pool area into 4 countries.
A further development of the EMPS model called SAMLAST (Hornnes, 1995) rep-
resents the transmission system between areas with a physical load flow model that sig-
27
nificantly improves the description of the power flow. Results can differ significantly
compared with a simple capacity constraints representation of transmission.
In the studies using NordMod-T, it was concluded that the construction of cables
were important for the results of CO2-emissions.
Kraftsim consider the total Nord Pool system as one area without any transmission
constraints between regions, except imports/exports to the continent.
In relation to the CO»-controversy, this simplification is justified by the fact that the re-
sulting price differences that occur between regions can be significant over short time in-
tervals, but are less significant (on average) in the long run.
Ongoing work at WSU has established a long-term system dynamics model of the West-
ern grid, including a 5-node power flow model (Dimitrovski et al. 2004) showing that it
is possible to represent the transmission system in a power marked system dynamics mod-
el.
Second, diurnal patterns and the dispatchability characteristics of generation technol-
ogies have been found to be important for the operations of transmission lines and should
thus be included in order to get a good picture of exchange between areas with different
characteristics. None of the models adequately represent dispatchability characteristics
of generation technologies.
8.3 Dispatchability features
Kahn etal. (1992) demonstrates that dispatchability features such as start-up and stop
costs are important for the economic profitability assessment of a project in a competitive
market. nuclear and coal can only slowly adjust generation and are thus run as baseload
units. Coal fired units would need 6 hours from cold start till max generation. Gas and
peak load turbines can adjust generation can quickly adjust generation and can be used for
load following.
In a detailed unit-commitment model, start-up and stop costs gives a more realistic
picture of the generation of each technology. Larsen (1996) used a detailed unit commit-
ment model of Preussenelektra (now a part of E-ON) to study the operational implications
of power exchange between the Norwegian hydropower system and Germany connected
through a transmission line.
The unit commitment model included start-up and shutdown costs for Preussenele-
ktras units. The results showed that power exchange between Norway (hydropower dom-
inated) and Germany (thermal dominated), will result in a shift towards higher utilisation
of baseload (coal) at the expense of medium- and peak load units (gas). The reason for
this is that coal units are cheaper in operation, but less flexible than medium- and peak
load units. Increasing power exchange with a hydropower system will then substitute
generation from some of the intermediate and peak load units during exports at peak hours
from Norway, and maintain an increased level of generation from coal during off peak
hours that can be exported and stored in the hydropower system.
Both EMPS and NordMod-T represent demand load in terms of load duration curves
(load blocks) which makes it difficult to incorporate start/stop costs that needs a chrono-
logical representation of load. Kraftsim on the other hand, has a chronological represen-
tation of load, but an hourly resolution with a description of start/stop costs of generation
units has not been implemented yet. Consequently, none of the models deal with technol-
ogy specific dispatch features that may be important for generation scheduling and con-
28
sequently CO5-emissions.
These shortcomings must be kept in mind when considering simulations involving
power exchange between hydropower dominated and thermal dominated systems.
[figure of price differences, Nord Pool Areas]
[figure of Nord Pool Spot price versus EEX spot price]
9 Discussion of modelling approaches
Good models are designed for specific purposes - huge amounts of time have been
devoted to developing such energy models. However, using models on problems outside
the scope of their original purpose inevitably cause omission of important cause-effect re-
lationships while disproportionately addressing others.
The EMPS model (originally developed for hydro scheduling and seasonal price
prognosis) only captured the short-term substitution effects, while investment substitution
effects were not discussed in the model studies.
Nordmod-T can in principle capture investment substitutions, but wind power was
exogenously represented in the simulation runs used for the analysis. Consequently, the
investment substitution effects were not sufficiently captured.
Kraftsim was particularly designed to analyse long-term versus short term implica-
tions of energy policies captured the both substitution effects. The model did not repre-
sent transmission constraints except for export/imports to the continent.
None of the models captured dispatchability features that are important for results on
power exchange between thermal and hydropower dominated systems. Including dis-
patchability features will most likely reduce the substitution effect of exchange to the con-
tinent, which was a major contributor to the results, particularly in the EMPS and the
NordMod-T study.
The modelling concept used here avoids this problem by being more of a flexible
modelling concept in which the model structure is tailored to the specific problem of in-
terest.
10 Conclusions
The results presented here shows that the fuel substitution strategy is a double-edged
sword. On one hand, substitutions in generation may reduce CO»-emissions. On the other
hand, investment substitutions may (in the Nordic case) substitute future investments of
renewables, and stimulate demand increases.
Could these results apply to other electricity markets than Nord Pool? Data used here
are specific for the Nordic countries, where renewables are becoming close to competitive
and environmental regulations are strictly enforced.
The short-term substitution effects depend on the short run marginal costs (SRMC)
of the technologies (i.e. SRMC supply curve), that can differ from country to country.
Nuclear and coal should not differ significantly between countries, the price of natural gas
may differ from country to country, although gas markets such as the EU market for gas
will in the long run reduce such price differences. The vintage of the production capacity
will also be of importance.
Concerning the investment substitution, this effect will heavily depend on the coun-
tries energy policy and availability of resources. The Nordic countries possess good wind
resources and wind energy is now close to competitive. In addition, renewables are sub-
29
sidised. This may not be the case in other countries with less renewable resource poten-
tial, natural gas is expensive, and coal may be an altemative for new investments.
But in many market where now renewables is a realistic option for investment, and
where coal is becoming less attractive due to CO2-quota obligations - this study warms of
the fuel substitution effect as being a counterproductive environmental policy as means of
reducing CO-emissions in the long run.
30
Notes
1, Seasonal price variations (Chapter 7.7.1)
A more precise estimation of water values will reduce seasonal price variations some-
what, and the model data needs to be improved in this respect. As the electricity market
become tighter, larger seasonal price variations can be observed. During the simulation
tun, the supply curve of generation technologies changes towards less peak load units and
less thermal baseload. The relative share of the flexible hydropower also diminishes, and
the share of wind power increase.
2. On boom/bust cycles (Chapter 7.7.1)
Potential boom and bust patterns in the electricity industry has been studied by Ford
(1999,2001) and Bunn and Larsen (1992). The underlying cause of the oscillations ap-
pearing in this study however, differs slightly from the previous studies. Firstly, acquisi-
tion of capacity in previous studies was determined by a demand forecast, where the
construction pipeline was taken into account to various degrees. Secondly, the models fo-
cused on capacity construction of mainly combined cycle gas turbines (CCGT), as they
are currently the cheapest technology for investments. In contrast, the simulation model
presented here, considers investments to be made purely on profitability criteria (for
which expectations of long-term electricity prices plays an important part, see chapter
7.2). Furthermore, there are nine different technologies to choose among, each with costs
changing in response to technology progress, price, fuel costs and resource availability,
and with different lead times in application processing and construction. Patterns of boom
and bust (shown as price oscillations) (compare LRMC’s in Figure 16).
A previous version of the Kraftsim model (V ogstad et al, 2003) with only one vintage,
and a fixed marginal cost curve for each technology did not exhibit similar patterns of
boom and bust. The model was however internally inconsistent since new investments
would alter the shape of the supply curve for each technology as new, more efficient
plants replaced old units.
31
References
Aune FR, T Bye and TA Johnsen 2000: “Gas power generation: Good or bad for the cli-
mate?” Revised version, Discussion paper no. 288, Oct 2000. Statistics Norway.
Available online www.ssb.no
Botterud A. Korpas M. Vogstad K-O. Wangensteen I. (2002): "A Dynamic Simulation
Model for Long-Term Analysis of the Power Market". Proceedings, Power Systems
Computation Conference 2002 Sevilla Spain.
BowerJ Bunn DW. Wattendrup C. (2001): “A model-based analysis of strategic con-
solidation in the German electricity industry” Energy Policy 29 pp 981-1005.
Dimitrovski A, M Gebremicael, K Tomsovic, A Ford and K Vogstad, "Comprehensive
Long Term Modeling of the Dynamics of Investment and Growth in Electric Power
Systems". 2004 EPNES Workshop, Mayaguez, Puerto Rico, July 13-14 2004.
Available [online] http://tomsovic.eecs.wsu.edu/V itae/Publications/DIM104c.pdf
Ford A, 1999: Cycles in competitive electricity markets: a simulation study of the western
United States. Energy Policy 27(11)
Ford A, 2001: Waiting for the Boom: A Simulation Study of Power Plant Construction in
California. Energy Policy 29(11)
Forrester JW. 1961: “Industrial Dynamics” MIT Press Cambridge MA
Hansen JV, J Hauch and MT Kromann (2001) “Will the Nordic Power Market Remain
competitive?” Working paper 2001:7, Det okonomiske rad, Denmark. Available
[online] http://www.dors.dk/arbpap/dors112.htm
Hornnes KS (1995): “ A model for coordinated utilization of production and transmission
facilities in a power system dominated by hydropower”. PhD dissertation, Dept of
Electrical Engineering, NTNU. ISBN 82-7119-865-3
IEA 2000 : Experience curves for energy technology policy. IEA 2000
Kahn E, C Mammay and D Berman, 1992: Evaluating dispatchability features in competi-
tive bidding. IEEE Transactions on Power Systems 7(3).
Larsen, TJ, (1996): Power Exchange between Norway and Germany: An Analysis of Op-
erational and Economic Consequences of Power Exchange between Statkraft and
PreussenElektra. MSc thesis, Dept of Electrical Engineering, NTNU.
Midttun A. Bakken B.E. and Wenstop F. 1996. Price formation and market stability
under different behavioural assumptions: Theoretical reflections underpinned by
computer simulation of liberal free trade in the Norwegian electricity market.
Morthorst P.E. 1999. “Capacity development and profitability of wind turbines” Energy
Policy 27 pp779-787.
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.
Nordpool 2001. Elbgrsen Market Report 4/2001. Available [online] 09.04.2002 http://
www.nordpool.no/
32
NVE 2002: Kostnader ved produksjon av kraft og varme i 2002. Handbook-2002, Nor-
wegian Water Resources and Energy Directorate.
SOU 2001: “ Handel med elcertifikat - ett nytt satt att framja el fran fornybara energikal-
lor” Svenska offentliga utredning.
Sterman JD. 2000: “Business Dynamics: Systems Thinking and Modeling fora Complex
World” McGraw-Hill. Book website: www.mhhe.com/sterman
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
Vector: http:// www.vector.no
Vogstad K, Belsnes M.M. Tande J.0.G. Homnes K.S. Warland G. (2001): Integras-
jon av vindkraft i det norske kraftsystemet. Sintef TR A5447 EBL-K 32-2001
Vogstad K. (2000) Utilising the complementary charateristics of wind power and hydro-
power through coordinated hydro production scheduling using the EMPS model.
Proceedings Nordic Wind Power Conference March 2000 Trondheim Norway.
Vogstad K (2004): A system dynamics analysis of the Nordic electricity market : The
transition from fossil fuelled towards a renewable electricity supply within a liber-
alised electricity market. PhD dissertation (forthcoming) available [online]
www.stud.ntnu.no/~klausv
Vogstad K, A Botterud, KM Maribu and S Grenaa (2002): The transition from a fossil
fuelled towards a renewable power supply in a deregulated electricity market. Pro-
ceedings, System Dynamics Conference 28th -1st August, 2002, Palermo, Italy.
Vogstad K, IS Kristensen and O Wolfgang (2003): Tradable green certificates: The dy-
namics of coupled electricity markets. Proceedings, System Dynamics Conference
July 20 -24. New Y ork, USA.
Wangensteen I. Botterud A. Grinden B. (1999): “Power exchange under various tech-
nical economical and institutional conditions” Sintef TR A5015 (Available in Nor-
wegian only)
33
Appendix 1 Data on existing capacity and marginal costs for the Nordic
power market.
ref1999 CO2avgitt 125
{krrico2}
Utslippsfak CO2-Produksjo
tor CO2 avgift_nskostnad
inkl. CO2-
avgfit
Type Produksjons Navn Brensel = MW kWh GWh — [tCO2/GW [krIMW kr/MWh tot
or profil malt hell hy
Nort-Sverige
20 refer kvo Fjemvarme ole 54 107 35044 150
21 refer kub Fjemvarme bio 481702
22 refer kvb Fjemvarme2 bio “10 7
30 koo Kondens ole 10 250 70088 340
31 koob Kondens ole 10 250 70088 340
35 gtod Gassturbin gassidies 8 420 1000 125 550
36 gtod Gassturbin2 gassidies 7 420 1000 125 550
TST
Syd-Sverige
20 varme Febo Indust Bioroye ear 554500 700
9 refer kj Kjemekraft kiemematr 100527071258 (vurdert)
30 refer kvk Kraftvarme kul 642 901210 820 103 193
31 refer kuko Kraftvarme kul/oje 641 1001048 70088 188
32 refer kug Kraftvarme ng 292 100466 40050 150
33 refer kvo Kraftvarme2 ole 188 «110488 650 Bl 191
34 refer kukb Kraftvarme2 kullibio 251304 40050 180
35 refer kub Kraftvarme2 bio 167170319 0
40 varme koob Kondens oljerbio 415250200, 700
45 varme gtod Gassturbin gassidies 180 420 10 1000125 545
Totale BEER] 73913
CO2avgitt 125
Utslippsfaktor CO2-avgift Produksjonsk
co2 ostnad ink.
CO2-avgfit
Typenr Type Navn Brens MW koMWh GWhEI GWh — {{CO2/GWhel} [kriMWh] —kr/MWh tot
al varme
Tylland og Fyn (DANME
VEST)
50 refer Vindkraft 1105 Prioritert 2050
20 refer Desentral kraftvarme 1374 Prioritert 6000 500 68
21 varme Deponigass gass 44 Priortert = 205. S11 431 54
2D refer kvk Esbjzerg kul 616107, 21651272 789 99 206
23 refer kvk Studsrup kul 700 «11831032629 854 107 224
24 varme kvk Vendsyssel kul 681 119.1565 4a 883 10 230
25 refer kvkFynsverket kul 6731192318 2735 866 108 27
26 varme kvk Ensted kul 6331154533258 849 106 22
27 refer _kvg Skaerbaek ng 400 «1551496831 450 56 22
Subtotal 6226 73035 8681
21385
Sjeetiand (DANM-
OST)
Teter Vindkraft 321 Promtert 550
20 refer Desentral Kraftvarme 466 Prioritert 2000 500 68
25 refer kvko Avedore 250 «11315961769 833 104 27
26 refer _kvko Amager 522 12122952738 865 108 229
27 refer kvko Aasnes 13821205356 S11 800 100 20
28 varme_kvko Stignes 4131281247 3 931 16 244
29 refer kvk Ostkraft 97 = 166. 99102 114g 14a 309
30 refer kvgo H.C, Orsted 290-209-337 1531 587 B 282
31 refer kg Svanemalle 1662182891184 508 64 282
35 varme gto Masneds 7 355 0 1288 161 516
36 varme kyo Kyndby 6139829 0 5062 633 2029
7608 13798 7831
Back to the Top
CC BY-NC-SA 4.0