June 17, 2003
MINATOM EXPORT CONTROL BEHAVIOR:
ECONOMIC FACTORS & MOTIVATIONS
Michael T. Maloney & Oana Diaconu
John E. Walker Department of Economics
Clemson University
Second Report to:
Lawrence Livermore National Laboratories
Sub-Contract B529375
COSTS ANALYSES AND METHODOLOGY RELATED TO NUCLEAR ELECTRIC
GENERATION
1. Introduction
The purpose of this research is to model the economics of the commercial nuclear
activities of Minatom. In our first report, we outlined the various commercial nuclear
export activities of Minatom with some commentary about the economics. In our
discussions with LLNL and with DOE, it was decided that we should focus on nuclear
generation activities. This report gives a summary of our analysis on two points
concerning nuclear electric generation: capital cost estimation and infrastructure factors
affecting levelized cost estimates.
As we noted in our last report, Minatom’s revenues from nuclear construction
projects abroad amounted to $424 million in 2001, about 17 percent of Minatom’s export
income. They were the third biggest source of export revenues after uranium products
and services (49.7%) and exports of nuclear fuel assemblies (18.7%).1 The forecast for
2002 was $900 million.
However, most of the “Atomstroyexport” nuclear construction abroad (Iran, India
and China) is “still accomplished through the credits given by the RF Government (the
repayment of which from the customers is either delayed or bounded by particular
additional conditions).”2 The state budget funding for these activities was $140 millions
in 2001 and another $250 million was budgeted for 2002.3
Minatom has commissioned power plants, research reactors and other nuclear
objectives in Bulgaria, former Eastern Germany, Finland, Hungary, Slovakia, Czech
Republic and China.4 It started a nuclear reactor in Cuba, at Jurangua, in 1983, but the
project was stopped in 1992 due to lack of funds and finally cancelled in 2001.The
1
International Business Relations Corporation, Department of Nuclear Energy and Nuclear Fuel
Cycle, Foreign Trade Policy and Authorized foreign trade entities of the Ministry of Russian Federation
for Atomic Energy, Annual Analytical Survey, Issue #1, Moscow 2003, pages 7 and 17-21.
2
Ibid., p. 87. Emphasis added.
3
See Maloney & Diaconu, “Is Nuclear Power Viable in Russia,” Electricity Journal, JanuaryFebruary, 2003, pp. 80-87, and “Commercial Initiatives in the Russian Nuclear Sector & the Nuclear
Nonproliferation Interests of the United States,” The Nonproliferation Review, Summer 2003.
4
Several former Soviet Union republics have Soviet design nuclear reactors: Armenia, Ukraine,
Lithuania, Armenia and Kazakhstan. They do not show up as being constructed by Russia, however.
Contact Information: Mailing: 223 Sirrine Hall, Clemson University, Clemson SC 29634-1309; phone: 864.656.3430;
email: maloney@clemson.edu; web: <www.clemson.edu/~maloney>
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
facility was more than 50percentage complete.5 It began the construction of 4 VVER 440
nuclear reactors at Zarnowiec in Poland but the construction was stopped in 1990. It has
conducted feasibility studies and design development for a nuclear desalination and
power plant in Lybia and for a nuclear power plant in North Korea, but none of the
projects have been carried on. Currently it has nuclear reactor business going on in
Slovakia, China, Iran, India and Ukraine and has plans to expand its business in
Kazakhstan, Syria, Burma, Vietnam, Egypt and eventually Turkey.
We see two major questions in the nuclear construction business initiatives of
Minatom:
1) Is it possible that Minatom is making money on nuclear power plant
construction given the price that it is charging for this work?
2) Are countries investing in nuclear power making economically efficient
decisions?
In this report we offer analysis of two points that contribute to our understanding of these
questions. Here we review and compare the capital cost estimates of nuclear power that
have been reported from many different sources. We also analyze country-specific
infrastructure that contributes to the efficiency of nuclear power.
The basic conclusions reached in this report are the following:
A) There is a great deal of variance in the forecast capital cost of nuclear power
plants. This is true not just for Russian construction but also for costs of other
reactors such as the Canadian and French models. Nonetheless, it is clear that
Russian units are typically forecast to have lower costs than the rest of the
competition, suggesting that Russia is consistently low-bidding. This
difference is somewhere between 15 and 25 percent, which is approximately
consistent with the notion that Minatom does not price design costs into its
bids and accounts only minimally for contingencies and interest during
construction.
B) Cost estimates such as those reported in a Finish study favorably comparing
nuclear power to other generation are based on assumptions that have to be
considered unattainable for many, and maybe even most countries. Electricity
cost per kWh is inversely proportional to load factor. The historical, worldwide average load factor for nuclear power plants is around 70 percent. Based
on the observed operation of their electricity systems, none of the countries
that are considering introduction of nuclear technology (in particular, Iran,
Egypt, and Turkey) can hope to do better than this average load factor for
nuclear power plants. This makes nuclear power very likely to be an
uneconomic source of generation.
Dalia Acosta, “Cuba Cancels Nuclear Plant Construction,” Tierramerica, archivo,
<www.tierramerica.net/2001/0211/acent.shtml> and Jonathan Benjamin Alvarado, “Nonissue: Cuba's
Mothballed Nuclear Power Plant,” July 1998, Center for International Policy, Washington D.C.
5
2
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
2. Capital Costs of Nuclear Power Generation
A nuclear power generator incurs three types of costs: capital costs (construction
costs including research development), operating costs (fuel and operation and
maintenance), and decommissioning costs. The storage and/or processing of the spent
fuel can be included in either decommissioning or fuel costs.
Nuclear capital generation has high capital costs and low generation costs.
Expressed as $/kWh capital costs represent about two thirds of nuclear energy costs while
fuel and other operating expenses are about one third.6 For coal fired and gas generation,
on contrast, fuel accounts for more than one half of the generation costs and capital cost
accounts for about one third. Thus, the competitiveness of nuclear power generation
relative to fossil fuel generation rests decidedly on how high nuclear construction costs
are relative to the prices of fossil fuels.
The capital costs of nuclear generation units are comprised of land acquisition and
site preparation, equipment and installation, interest cost during the construction and
nuclear decommissioning plant costs. Contingencies for cost overruns are also included.
A typical breakdown of estimated nuclear power capital cost for a pressurized
water reactor is shown in Table 1.
Obviously the choice of the discount rate and the construction time will affect the
total capital cost of the plant through the interest during construction component. The
typical construction cost for a nuclear unit is 5-9 years. Longer construction times caused
by regulatory delays have often been blamed for the escalation of construction costs of
nuclear power plants in the past two decades, and are largely responsible for the demise
of nuclear power in the United States.
Table 1. Break Down of Construction Costs for
Nuclear Generator
Item
Percent of Cost
Civil engineering and building
21%
Design and consultancy cost
15%
Nuclear steam supply system
13%
Turbine generator
12%
Other electrical plant
9%
Other mechanical plant
8%
Contingencies
Total Excluding Interest
Interest during construction
22%
100%
28%
Source: Thomas (1988), p. 40.
6
S.D. Thomas, The realities of nuclear power, Cambridge University Press, 1988, p. 39. Estimates
from Diaconu & Maloney (2002) find the ratio of capital costs to total for nuclear power in Russia to be 78
percent and 67 percent in the United States. The U.S. estimate is based on the construction costs of pre1970 nuclear power plants, inflated to 2002 dollars.
3
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
The estimates are also sensitive to the discount rate used. In the international
practice 10 percent is considered a reasonable rate although for some developing
countries a higher rate might be appropriate.7 For the purpose of the comparison
presented in this study, we will use 10 percent across the board. At a 10 percent discount
rate, interest cost during construction represents 25-30 percent of the total construction
cost of a nuclear unit.8
Although not included in the table above, the decommissioning costs are typically
included in capital cost estimates. Decommissioning cost amounts to 2 percent of the
cost. Some studies apply a lower discount rate for decommissioning and waste
management reflecting a long run rate of return expected to be earned on the funds set
aside for these purposes. For most of the data coming from the Organization for
Economic Cooperation and Development (OECD) study (see below), the
decommissioning and waste management have been discounted at the interest rate.
Particular to nuclear power plant construction, cost structure is a high level of
constant costs, independent of plant size or life of the plant, for the project
implementation. Activities, such as site approval, information and public relations,
training the operation and maintenance personnel, licensing and activities of the supplier
such as safety analysis, documentation and project management, can represent as much as
30 percent of the total capital cost of a one unit nuclear power plant of 1200MW
according to some sources.9 However, this percentage seems exaggerated based on U.S.
experience.
Construction cost is related with the number of units to be built on a site. “The
cost reduction is calculated to be 10 percent for a site with two units and 20 percent for a
site having 4 units provided that the construction interval is less than 2 years.”10 A similar
phenomenon exists for scale effects in the size of the unit. That is, there is roughly a 20
percent cost saving as plant size increases. This estimate is based on the scale factors
observed in U.S. construction costs across all types of generators.
2.1.
Data on Capital Cost of Nuclear Power Generation
Much of the data on capital cost of nuclear power generation comes from the
United States. Capital cost expenditures are available for existing plants on FERC Form 1
distributed by the U.S Department of Energy. We used these data extensively in our early
research. Outside the United States, estimates of the capital costs are much harder to
7
The methodology of choosing the discount rate is ill-defined in policy discussions. We use a
financial economics approach wherein the discount rate is based on the business risk of the venture. While
it may be difficult to precisely determine the inherent business risk, 10 percent seems reasonable and 5
percent is highly questionable. In the calculation of capital cost, the interest rate determines the cost of
funds during construction. In the calculation of levelized cost, the interest rate is the discounting factor for
revenues throughout the life of the plant.
8
Bertel, Evelyne, and Geoffrey H.Stevens, Comparative Costs of Generating Electricity, OECD
Nuclear Energy Agency, France, at 11th Pacific Basin Nuclear Conference (PBNC), Banff, 1998.
9
Ferroni, Ferrucia, Hans-Jürgen Kirchhof, and Juan B. Heredia, Review of Cost Reduction
Measures for Nuclear Electricity, Electrowatt Engineering Ltd, Switzerlant, PBNC, 1998.
10
Ferroni, et al., (1998).
4
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
come by and are much less detailed. With few exceptions we were unable to find unit
level open source data on capital costs internationally.
Data are available in a study done by the OECD on the cost of electricity across
countries for all types of generation.11 Their data was obtained by circulating a
questionnaire to OECD member countries and non-OECD participant countries through
the International Atomic Energy Agency (IAEA). The study is comprehensive and has
been updated several times.
The following quote summarized the scope of the study and the definition of
some of the terms:
The questionnaire asked for information on all types of base-load power plants expected
to be commercially available for commissioning by 2005-2010, except hydropower
plants (...) The technologies for which information could be provided included, primarily,
state of the-of-the art power plants (category A) as their performance and costs, based
upon ordered plant prices, quotation or detailed paper analysis are considered to be well
established. Cost information was sought also on technologies under development that
are expected to be commercially available by 2010 (category B plants)(…) Thirteen
countries provided cost estimates for nuclear power plants including three reactor types:
pressurized water reactors (PWR); boiling water reactors (BWR); and pressurized heavy
water reactors (PHWR). The size of the nuclear units ranges from 455 MWe to 1460
MWe. Three countries, France, Japan and China, estimated nuclear fuel cycle costs
corresponding to a closed cycle with reprocessing and recycling, while the other
countries provided costs estimates corresponding to an open cycle with direct disposal of
spent fuel.12
Table 2 summarizes the data found in the OECD study. The data provided in the
study are rather generic. Although the size and the technology on which the estimate is
based are specified, the name of the power plant or of the site is not. In part, this is due to
the fact that some of the estimates assume a construction program of several reactors
(e.g., France, 10 units; Korea, 6; and Russia, 5). Another reason for scarcity of plant data
is that, in many countries, especially those that have started the process of liberalizing the
power sector, capital cost information is considered commercially sensitive.
The only specification about the source of the estimate is whether the estimate is
based on the ordered plant (O), on a planned plant (P) for which a paper study exist, if the
source is a quotation (Q), or on a feasibility study (FQ). For some countries, several
characteristics of the plant (capacity and number of units on site) allow us to guess the
name of the actual power plant on which the estimation was based but such identification
is speculative. For the estimates for which the number of units on site were provided we
assume that the economies of multi-unit siting were taken into account.
Besides the OECD/IEA study, there are several studies on specific countries that
provide some information on the domestic capital cost of nuclear generation.
Unfortunately, again, for most of them plant level data is missing. We identified the units
to the extent it was possible. Many of the studies actually base their estimates on OECD
data. These studies are referenced in the notes of Table 3.
11
Organization for Economic Cooperation and Development (Paris), Nuclear Energy Agency,
International Energy Agency, Projected Costs of Generating Electricity, Paris, 1998.
12
OECD/IEA (1998), p. 22.
5
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
We pulled together the data from all of these sources and transformed the
estimates into 2002 U.S. dollars. Whenever we have multiple sources, we try to discuss
the differences between estimates. However there are several reasons for cautions in
comparing the country data.
As the authors of the OECD/IEA (1998) remarked, transforming data into U.S.
dollars using the existing exchange rate may seriously distort the estimates, especially the
one referring to non-OECD, developing countries. 13 Exchange rates do not necessarily
reflect purchasing power parity. That is, purely domestic construction (steel, concrete,
labor, etc.) in one country may be identical to that in another country, but the exchange
rates may be different. This means that a foreign investor seeking to build plants in the
two countries would face different costs, but domestic investors in the separate countries
would face identical costs.14 It is also true that exchange rates may be distorted by
currency controls and fixed exchange rates that are not supportable in real transactions.
A number of other factors besides the imperfect conversion in a common unit of
currency may cause the estimates to vary both across, within countries and among
projects such as:15
a) Differences in technology. As Thomas (1988) pointed out, for example, “in the
UK the nuclear steam supply system for an AGR was estimated to cost nearly
four times as much per kilowatt as that required for a PWR.”16
b) Differing cost components. Countries that have mass-production facilities and
highly standardized reactors, like France, may be able to manufacture the reactor
components cheaper.
c) Differences in regulatory requirements.
d) Economies of scale and multi-unit siting not accounted for in the data.
Many of the estimates on the capital cost of nuclear power plants have proved to
be underestimates. Caution should be exercised especially in judging the data based on
planned or under construction plants.
13
OECD/IEA (1998).
We do not think of this as a serious issue because the fact is that with nuclear power, there is
almost always international investment. Moreover, the main point of this research is to investigate the
economics of Russian nuclear export activities. Exchange rates and not purchasing power parity is the main
driver.
15
See Thomas (1988), p. 39.
16
Ibid., p. 39.
14
6
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
Table 2. Nuclear Power Plant Specifications in OECD/IEA Study
Country
Reactor
Type / Fuel
Cycle
Option
Units
in
Project
Canada
PHWR/OT
Total
Units
at
Site
Net
Capacity
(MW)
Net
Thermal
Eficiency
(%)
Cooling
Tower
Site
Cost
Estimation
Source
/ Date
2
665
31.2
No
New
O/96
PHWR/OT
2
881
31.2
No
New
Q/96
Finland
BWR/OT
1
1000
33.0
No
New
P/96
France
PWR/CC
1
4
1460
33.0
Yes
New
Q/96
Japan
ABWR/CC
1
4
1303
33.0
No
New
P/96
Korea
PWR/OT
1
2
1000
35.1
No
New
O/95
Spain
PWR/OT
1
1000
34.0
Yes
New
P/96
Turkey
PWR/OT
1
1000
34.5
Yes
New
P/95
Brazil
PWR/OT
1
1229
34.6
No
Existing
Q/95
China
PWR/CC
2
592
29.6
No
New
P/96
PWR/CC
2
935
32.2
New
FS/96
PHWR/OT
2
665
29.1
No
New
FS/96
India
PHWR/OT
2
455
29.0
No
Existing
P&O/97
Romania
PHWR/OT
1
5
707
31.0
No
Existing
P&O/95-97
Russia
VVER/OT
1
5
604
33.3
No
Existing
Q/96
United
States*
PWR/OT
1
1300
32.0
Yes
New
P/96
Brazil*
PWR/OT
1
1229
35.0
No
New
P/96
3
Notes: * Category B technologies: under development and projected to be available by 2010. Reactor types: PWR pressurized light
water reactor; PHWR, pressurized heavy water reactor; VVER, Russian pressurized light water reactors; BWR, boiling water reactor;
ABWR, advanced boiling water reactor. OT and CC referrers to the fuel cycle option chosen. OT is an open fuel cycle or “once
through” and CC means “closed cycle”. “Units in project” column refers to the number of units included in the cost estimate.
Source: OECD/IEA (1998), p. 46.
2.2.
Comparisons of Capital Costs Estimates for Nuclear Generation
Capital cost estimates from all available open sources are presented in Table 3.17
The cost estimates represent all expenditures and encumbrances at the start of
commercial operation, including interest cost during construction. Where the source gave
17
Because the data includes only the most recent available estimates on nuclear generation capital
costs there are some notable absence in the data. Germany, for example, had a quite successful nuclear
program that was halted in the late 80’s (last reactor, NECKARWESTHEIM-2 came on line in 1989). Data
on the cost of the latest nuclear German reactors is not available.
7
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
a construction only or so-called overnight cost, we added expected construction interest
and decommissioning, usually using the OECD assumptions as discussed earlier. The
numbers are in 2002 U.S. dollars.
Overall the database contains 31 cost estimates from different sources on 15
countries.18 The most detailed data available is on China where we have 9 cost estimates
from 3 sources.
The capital costs vary between $1500/kW (an estimate for a Russian VVER in
India, at Kudankulam) and $3768/kW (our estimate for a U.S. PWR) with an average at
about $2500/kW. Net capacity varies between 455MW and 1460MW per unit.
Most of our estimates are on pressurized light water reactors PWR (14), followed
by pressurized heavy water reactors PHWR (9, mostly Canadian), Russian pressurized
light water reactors VVER (5), one boiling water reactor BWR and one advanced boiling
water reactor ABWR. BWRs are the most expensive reactors in our sample with an
average cost $3176/kW, but there are only three estimates and the average is driven by
the very expensive Japanese ABWR. The least expensive are the Russian VVER (average
$1988/kW) constructed domestically or abroad (in China and India) although the
estimates for domestic construction are significantly higher than the estimates for reactors
abroad. The OECD estimate of the domestic Russia units is significantly higher than
other estimates. The average for PWR units is $2656/kW. This falls to $2570/kW if the
highest estimate (based on early U.S. experience) is thrown out. The average for PHWR,
which are mostly Canadian construction, is $2457/kW.
Countries with the lowest capital cost per kW (first quartile) are China, with an
average capital cost of $2238/kWh and India with $2136/kW. However the averages are
mostly drawn by the very low cost of the reactors built in both countries by Russia. In
China, if we drop the estimates for the Russian reactors, the country average capital cost
becomes $2383/kW. The country average for imported technologies (4 French reactors
and 2 Canadians) is $2467/kW, very close to the sample average. In India, the average
without the Russian built reactors is $2454, again close to the sample average.
The most expensive reactors (last quartile) are in United States, Japan, Spain, UK
and Finland. The numbers range between $2871/kW, the OECD estimate for a planned
Finnish reactor, and $3768, our estimate extrapolated from the early United States
experience.
Right in the middle are the estimates for French and Korean reactors and for two
Canadian built Chinese reactors. Brazil and Argentina estimates are also very close to
average but the cost for Argentina is most likely grossly underestimated as the actual cost
for the specific plant itself.19
For both Canadian and French reactors, reported costs for domestic reactors are
similar to those of exported reactors. Canada, for example, has a capital cost of
$2455/kW for its domestic reactors, and about $2400/kW for its exported reactors (to
China and Romania). France’s estimates for domestic construction are $2523/kW while
18
There is one plant for which there are two separate cost estimates.
The construction of Atucha 2 started in 1981 and has been halted several times since. It is
uncertain when and whether the reactor will be finished. For a reactor with a construction period that
stretched over more than 20 years the estimate we have is unrealistically low. The estimate was obtained by
adding 35 percent for interest during construction and decommissioning to an overnight capital cost
estimate of $1800/kW provided by EPOA (2000).
19
8
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
the reactors exported to China cost $2385/kW on average. The lower cost of the French
technology in China could be due to the fact that China typically undertakes part of the
reactor construction.20 For Russia, on the other hand, the domestic capital cost is at least
$2266/kW while the exported reactors average $1665/kW.
For the countries on which we have estimates from different sources, the
estimates do not converge. China’s estimates vary between $1620/kW for the 1000 MW
Russian VVER to $2747/kW for the 2 unit 665 MW PHWR.21 The discounts at which
Russia seems to sell its reactors abroad only partially accounts for the low cost of the 2
VVER reactors. Russia will receive $2.4 billion for the two units. While Russia is
supplying the technology and the reactor itself, China is responsible for construction and
installation. Russia is financing the $1.3 billion loan at 4 percent. Based on our estimates,
Russia may be enjoying a modest profit on its construction activities in China eroded,
however, by the financial obligations of the contract.22 If China is able to supply such a
significant share of the construction at such low cost, the question is why it does not do it
for the rest of reactors it buys from abroad? It may be that the significantly higher capital
costs of the Canadian and French reactors include a payment for the technology transfer
while the Russian price does not.
The estimates vary even more for India. The cost of the two 1000 MW VVER
reactors that Russia is building at Kudankulam is as low as $1500/kW. This number is
based on news reports that place the total cost of the project at $3 billion. It is probably
appropriate to discount this estimate because we do not know what it comprises (e.g.,
construction interest may be excluded). The OECD country estimate for PWHR reactors
is $2719/kW while another country estimate places the capital cost at $2189/kW.23 It is
obvious that Russia is selling the VVER reactors at a significant discount, based both on
the capital cost estimates for VVERs in Russia and based on the competitive supply
prices of other nuclear units.
Notable also is the difference between OECD and WNA estimates in Finland.
OECD estimates that the capital cost of BWR technology in Finland is $2871/kW while a
WNA country estimate of capital cost (including the initial fuel load!) is $1637/kW.24
The same study however states that the 1000MW reactor planned to be built in Finland in
the next few years is going to cost between $1870/kW and $2670/kW.25 We have no
20
The financing cost of the reactor in developing countries should be higher, however than for the
domestic reactors. Both France and Canada have a history of undertaking debatable financing arrangements
with developing countries. For an interesting account of these issues see, for example, “The Candu
Syndrome” by David H. Martin, Nuclear Awareness Project for the Campaign for Nuclear Phase Out,
1997, at <www.ccnr.org/turkey_syndrome.html>.
21
As estimated by CEPO(1998).
22
See Maloney & Diaconu (2002).
23
EPOI (1998) gives only the base cost. We obtained the total capital cost by assuming that
interest during construction and decommissioning represent 25 percent of the base cost as they are in the
IEA/OECD study.
24
Risto Tarjanne & Sauki Rissanen “Nuclear Power: Least Cost Option for Baseload Electricity
in Finland”, The Uranium Institute Twenty Fifth Annual International Symposium 2000, <www.worldnuclear.org>, Table 1, p. 8. The paper states that the capital cost of a nuclear plant in Finland is 1749 Euros
(2000) per kW. We have transformed it into 2000 U.S. dollars assuming as the authors do an exchange rate
1 Euro=$.90. Then we inflated the number to 2002 U.S. dollars.
25
The reactor is expected to cost between 1.75 and 2.5 billion euros in 2000 prices. Capacity of
the reactor is not yet decided. It depends on the reactor that will be selected. "In March 2003 tenders were
9
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
explanation for the extremely low number in the study cited by WNA. It does include
construction interest.
Table 3. Capital Costs of Nuclear Power Generation
Country & Plant
Net
Name
Source Estimate
Capacity
Technology
$ / kW
Operational
CHINA
Daya Bay (1&2)
CEPO (1998)1
944
Qinshan II
IEA/OEC (1998)2
610
Qinshan III (1&2)
CEPO (1998)
665
Qinshan III (1&2)
665
Qinshan III (1&2)
IEA/OECD (1998)
WNA contract value
(1996)3
Lingao (1&2)
CEPO (1998)
930
Lingao (1&2)
Tianwan 1&2
Tianwan 1&2
IEA/OECD (1998)
CEPO (1998)4
NTI(2001)5
930
1000
1000
French
PWR
Chinese
PWR
Canadian
PHWR
Canadian
PHWR
Canadian
PHWR
French
PWR
French
PWR
VVER
VVER
INDIA
n/a
Tarapur 3&4
Kudankulam 1&2
EPOI (1998)6
IEA/OECD (1998)
M&D (2003)7
n/a
455
1000
IEA/OECD (1998) &
EPOK (1999)8
FINLAND
n/a
n/a
Loviisa/Olkiluoto
2285
1994
1872
2002
2747
2002
2402
2002
2496
2002
2182
2002, 2003
2691
1875
1640
2002, 2004
2004, 2005
2004, 2006
PHWR
PHWR
VVER
2189
2719
1500
n/a
2006, 2007
2007, 2008
1000
PWR
2501
2002
IEA/OECD (1998)
WNA (2003)9
WNA (2003)10
1000
n/a
1000
BWR
n/a
n/a
2871
1637
2270
2005-2010
2005-2010
2007
RUSSIA
n/a
n/a
IEA/OECD (1998)
D&M (2003)11
604
1000
VVER
VVER
2709
2266
n/a
n/a
UNITED STATES
n/a
n/a
M&D (2003)
IEA/OECD (1998)
1000
1300
PWR
PWR
3768
2285
n/a
n/a
665
KOREA
Yonggwang 5&6
submitted by three vendors for four designs: Framatome ANP: European Pressurized Water Reactor (EPR)
of 1500 MWe and the SWR-1000 (a BWR) of 1200 MWe, General Electric: European Simplified Boiling
Water Reactor (ESBWR) of 1390 MWe, and Atomstroyexport: VVER-91/99 of 1060 MWe.
(Westinghouse did not bid its AP-1000 PWR or its BWR-90+) The EPR is the new standard design for
France, the two BWRs are both undergoing design certification in the United States, and two of the VVER91 units are being built in China" WNA, Nuclear Power in Finland, April 2003.
10
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
FRANCE
n/a
IEA/OECD (1998)
1460
PWR
2523
n/a
CANADA
n/a (2 units)
Darlington (1&2)
IEA/OECD (1998)
IEA/OECD (1998)
665
881
PHWR
PHWR
2638
2272
n/a
1990, 1992
JAPAN
Shika-2
IEA/OECD (1998)
1303
ABWR
3481
2006
IEA/OECD (1998) and
EPOB (2000)12
1229
PWR
2610
2001
745
PWR
2475
n/a
BRAZIL
Angra 2
ARGENTINA
Atucha 2
EPOA (2000)13
SPAIN
n/a
IEA/OECD (1998)
1000
PWR
3272
n/a
TURKEY
n/a
IEA/OECD (1998)
1000
PWR
2600
n/a
ROMANIA
Cernavoda 1
Cernavoda 2
IEA/OECD (1998)
n/a 14
707
707
PHWR
PHWR
2304
2348
1996
2007
GREAT BRITAIN
Sizewell B
Hinkley C
KU (1995) 15
KU (1995)
1155
1175
PWR
PWR
3097
3028
1995
n/a
11
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
Notes:
Where plant names could be reasonably inferred, they are listed.
1. William Chandler, Guo Yuan, Jeffrey Logan, Shi Yingyi, Zhou Dai "China's Electric Power Options. An Analysis of
Economic and Environmental Costs,” [CEPO] Advanced International Studies Unit, Pacific Northwest National Laboratory
[PNNL], June 1998 See p. 43 for capital cost estimate on China's Power Plants.
2. International Energy Agency (IEA), Nuclear Energy Agency and Organization for Economic Cooperation and
Development (OECD) "Projected costs of generating electricity"), Paris 1998, p. 54-55.
3. WNA News Briefings, 20-26 November 1996, <www.world-nuclear.org> citing Reuters, November 15, 1996.
4. Range: 1730 to 2020.
5. Russia: General Nuclear Exports Developments, November 28, 2001, <www.nti.org>.
6. Shukla, P.R., Debyany Ghosh, William Chandler, Jeffrey Logan, "Electric power options in India," [EPOI] Pew Center on
Global Climate Change, October 1999, p. 26.
7. Michael T. Maloney and Oana Diaconu , “Analysis of Privatization of Russia’s Nuclear Ministry on the Nuclear
Nonproliferation Objectives of the United States”, Special Report, Strong Thurmond Institute, 2001,
<www.clemson.edu/~maloney/papers/cudoereport.pdf>, citing news reports.
8. Jin-Gyu Oh, Jinwoo Kim, Jeffrey Logan, Sung Bong Yo, Willian Chandler, Dong-Seok Roh
"Electric Power Options in Korea", Pew Center on Global Climate Change, October 1999, p. 37 and OECD (1998). EPOK
uses the same estimate as OECD as the base construction cost. Consequently, I consider their overall capital cost estimate
identical (10% discount rate).
9. Risto Tarjanne & Sauki Rissanen “Nuclear Power: Least Cost Option for Baseload Electricity in Finland”, The Uranium
Institute Twenty Fifth Annual International Symposium 2000, <www.world-nuclear.org>, Table 1, p. 8.
10. Nuclear energy in Finland, April 2003, <www.world-nuclear.org>. Range: from 1870 to 2670.
11. Oana Diaconu and Michael T. Maloney “Is nuclear power viable in Russia?" The Electricity Journal, January/February
2003, p. 82. The estimate is based on OECD estimate scaled up to 1000MW, the size of the standard unit being built by
Russia today using an average cost elasticity of -0.2. That is for each 10 percent increase in size, capital cost per MW goes
down by 2 percent.
12. Roberto Schaeffer, Jeffrey Logan, Alexandre Salem Szklo, William Chandler, Joao Carlos de Souza Marques "Electric
Power Options in Brazil", Pew Center on Global Climate Change, May 2000.
The base capital cost in this study is $1600/kW (1997 $US), very close to IEA/OECD $1550/kW. We consider the overall
estimates for capital cost identical.
13. Daniel Bouillle, Hilda Dubrovsky, William Chandler, Jeffrey Logan, Fernando Groisman "Electric Power options in
Argentina", Pew Center on Global Climate Change, May 2000, p. 16.
14. This estimate was recorded in our database, but the source has been lost.
15. Anne Ku "Modeling Uncertainty in Electricity Capacity Planning", Unpublished Thesis, London Business School,
February 1995, p. 203. She cites two studies for the source of her data OECD/NEA (1989), "Projected Costs of generating
Electricity from Power Stations for Commissioning in the Period 1995-2000," for the estimate on Sizewell and OECD/NEA,
IEA and UNIPEDE (1988) "Electricity Generation Costs: Assessment Made in 1987 for Stations to be Commissioned in
1995", Sorento Congress, May 5- June 3 for the estimate on Hinkley.
2.3.
Implications of the Results
We draw the following conclusions from the comparison of the cost estimates
shown above. There is a great deal of variance in the forecast capital cost of nuclear
power plants. This is true not just for international Russian construction but also for costs
of other reactors such as the Canadian and French models. The French units have a mean
of $2386/kW with a standard deviation of $270; the Canadian models have a mean of
$2548/kW with a standard deviation of $178.
Nonetheless, it is clear that Russian units are typically forecast to have lower costs
than the rest of the competition. If we look only at the estimates for Russian units
installed in foreign countries, the average is $1672/kW. This is probably biased
downward because of the one observation for the Indian facilities based on simple news
reporting. Even so, if we go with the highest of these estimates, $1875/kW cited by
PNNL, the number is still 21 percent lower than the French cost and 26 percent lower
than the Canadian.
The suggestion is that Russia is consistently low-bidding. This difference is
somewhere between 15 and 25 percent. Based on the cost breakdown presented in Table
1, this range is roughly consistent with the notion that Minatom does not price design
costs into its bids and anticipates minimal contingencies.
12
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
3. Analysis of Load Factors at Nuclear Power Plants
The analysis presented in the last section also speaks to the issue of levelized cost.
Levelized cost is the calculation of the per kWh price that the power plant must receive
throughout its life in order to pay back the construction and operating cost. Levelized cost
is based on the capital cost of the machine as we have discussed in section 2 above. It
also depends on the operating costs (e.g., maintenance, fuel, etc.), the forecast life,
discount rate, and load factor.26
Table 4 shows a simple comparison of levelized cost estimates based on
alternative assumptions. In the first case we use assumptions that seem to us to reflect the
true state of operation of nuclear power plants. We call these the baseline assumptions.
We use a life of 30 years because historically, most plants have required substantial
capital improvements to continue operation past that point. We use a capital cost estimate
that is around the cost reported for French units. We price fuel at the cost paid by U.S.
utilities; the same is true for maintenance. Alternatively, we show the “best-case”
assumptions used in the Finish study.27 The effect on cost is dramatic. A conservative
estimate of the cost of nuclear generation is nearly three times higher than the cost
estimated on best-case scenarios.
Table 4. Levelized Cost Comparison for Nuclear Power
Assumptions
Factors:
Baseline
Best-Case
Plant Life (years)
30
40
Discount Rate
0.1
0.05
Load Factor
0.8
0.9
Capital Cost ($/kW)
2400
1637
Fuel cost ($/MWh)
6.0
2.9
Maintenance Cost ($/MWh)
9.0
3.4
Implied Cost
Levelized Capital Cost ($/MWh)
Full Cost of Power ($/MWh)
36.3
51.3
12.1
18.4
It is possible that a best-case scenario is appropriate for Finland. We know that
Finland has experienced favorable performance from its existing nuclear power plants
and that the electrical system overall operates efficiently. However, it is romantic to make
26
Load factor is the amount of power produced by a generator divided by the engineering capacity
of the unit. Usually load factors are stated for a year. The calculation, then, is the total kilowatt hours of
power generated by the unit divided by the capacity of the unit in kilowatts times the number of hours in
the year.
27
Risto Tarjanne & Sauki Rissanen “Nuclear Power: Least Cost Option for Baseload Electricity in
Finland”, The Uranium Institute Twently Fifth Annual International Symposium 2000, <www.worldnuclear.org>, Table 1, p. 8.
13
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
similarly sanguine estimates for less developed countries.28 In this section we offer an
analysis of one of the most important factors affecting levelized cost, load factor, and link
it to the operating efficiency observed for the electric system as a whole. From this
analysis we are able to make forecasts of the expected cost of nuclear power even for
those countries that have not yet adopted nuclear generation technology.
3.1.
Load Factor Effects
In our earlier work, we assumed a baseline scenario for load factors in Russia of
79 percent. However, nuclear power plants worldwide have not historically enjoyed load
factors this high, nor has Russia. Since levelized capital cost is inversely proportional to
load factor, the relationship between the two is a significant factor in determining the true
cost effectiveness of nuclear power. A 10 percent decrease in load factor, say, from 79
percent to 71 percent, increases capital cost by 10 percent.
The worldwide historical experience in load factor is 69.4 percent for reactors
currently operating and 68.3 percent for all commercial reactors.29 These numbers are
capacity weighted averages by machine by year. The following two tables show the
worldwide experience by year and by country. By year, load factors have been
improving. Load factors were 50 to 60 percent in the early 1970s. They have increased to
around 80 percent today. Even so, there is still a wide range of operating performance
across countries. Even looking at the most recent experience, countries such as Finland,
Belgium and Switzerland that operate in the high 80s and low 90s of load factor are offset
by countries like India which is in the 50s.
Many things can affect load factor for an electric generator. For nuclear plants, in
particular, age is important. Because of the complexity of the machinery and controls, it
is common for nuclear plants to operate at less than full power when they first come
online. Indeed, on average it takes nuclear power plants 9 months from first powering the
reactor until commercial operation, and 10 percent of the time it takes more than a year.
Even in the first year of operation, the reactor is will not run at full power or be
synchronized to the grid all the time. Age also works against reactors. On average, the
older they get, the lower their load factors.
We capture these many factors in the following statistical analysis. We use
multiple regression to model the load factor at each plant as a function of the
characteristics of that plant. Plant characteristics include the age of the plant and the
length of the construction period. We use {0,1} dummy variables to take account of the
period from the reactor startup and the first year of commercial operation, the first year of
commercial operation, and the last year of commercial operation for reactors that have
been shut down. In addition, we include dummy variables for each country and year pair.
28
As a side note on age, of the commercial nuclear power plants that operated for 5 or more years
and were subsequently shut down, the average age is 20 years, not 30, and the maximum life is 35.
Moreover, no nuclear power plant has yet enjoyed 40 years of service.
29
Data on design, construction, and operating characteristics of nuclear power plants worldwide is
available from the International Atomic Energy Agency (IAEA) through its Power Reactor Information
System (PRIS).
14
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
Table 5. Load Factors for Nuclear Power Plants—
World Wide Experience by Year
Year
All Reators
Currently Operating
1970
0.53
0.53
1971
0.56
0.56
1972
0.54
0.52
1973
0.55
0.53
1974
0.54
0.51
1975
0.58
0.56
1976
0.60
0.58
1977
0.59
0.58
1978
0.61
0.62
1979
0.58
0.59
1980
0.57
0.57
1981
0.59
0.59
1982
0.59
0.59
1983
0.61
0.62
1984
0.62
0.63
1985
0.66
0.66
1986
0.65
0.66
1987
0.64
0.67
1988
0.65
0.66
1989
0.65
0.67
1990
0.66
0.69
1991
0.68
0.70
1992
0.68
0.70
1993
0.70
0.71
1994
0.70
0.71
1995
0.71
0.73
1996
0.72
0.73
1997
0.71
0.73
1998
0.74
0.75
1999
0.77
0.77
2000
0.78
0.78
2001
0.80
0.80
2002
0.85
0.85
15
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
Table 6. Load Factors for Nuclear Power Plants—By Country
Country
All Reactors Currently Operating
Since1996
Armenia
0.52
0.54
0.53
Argentina
0.74
0.74
0.81
Belgium
0.82
0.82
0.89
Bulgaria
0.53
0.53
0.53
Brazil
0.35
0.35
0.55
Canada
0.70
0.76
0.77
Switzerland
0.84
0.84
0.88
China
0.73
0.73
0.76
Czech Republic
0.79
0.79
0.84
Germany
0.73
0.76
0.86
Spain
0.77
0.78
0.88
Finland
0.86
0.86
0.93
France
0.65
0.67
0.72
United Kingdom
0.67
0.66
0.74
Hungary
0.83
0.83
0.87
India
0.45
0.45
0.56
Italy
0.37
Japan
0.72
0.72
0.80
South Korea
0.82
0.82
0.86
Kazakhstan
0.20
Lithuania
0.48
0.48
0.43
Mexico
0.69
0.69
0.75
Netherland
0.79
0.79
0.92
Pakistan
0.29
0.29
0.34
Romania
0.75
0.75
0.75
Russia
0.64
0.64
0.62
Sweeden
0.71
0.71
0.76
Slovenia
0.72
0.72
0.83
Slovak Republic
0.71
0.71
0.69
Taiwan
0.79
0.79
0.83
Ukraine
0.62
0.63
0.68
United States
0.67
0.68
0.82
South Africa
0.62
0.62
0.78
As noted above, we expect operation before the plant enters its commercial phase
to be characterized by relatively low load factors. This same phenomenon is likely to be
true in the first and last years of commercial operation because the plant does not operate
for the whole year. Age, in and of itself, is likely to be negatively related to load factors
at least after some point. We allow for a varying age effect by including a squared term.
We also test to see if there is a size effect; that is, we check to see if big plants are
generally more efficient than small ones. We include the length of the construction period
on the hypothesis that plants delayed in construction are not likely to run as well as those
16
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
that are finished in a timely fashion.30 Finally, a dummy variable for plants that have been
shut down gives a differential in operating efficiency between operating and closed
facilities.
Table 7. Regression of Load Factor on Plant Characteristics
Independent Variables:
Coefficient
Age of the Power Plant
-0.004
Length of Construction Period
-0.009
Prior to Commercial Operation*
-0.351
First Year of Commercial Operation*
-0.200
Plant has been Shut Down*
-0.070
Last Year of Operation*
-0.255
F-stat
Classification Variables for Country and Year
22.81
R-squared
.491
No. of Observations
9331
t-stat
-8.31
-9.86
-24.33
-21.22
-10.36
-10.91
d.f.
(776, 9330)
Notes: (*) denotes {0,1} dummy variable. Age in years from time of commercial operation. “Prior to
Commercial Operation” is a dummy variable for years from reactor startup and time of commercial operation.
The estimates shown in Table 7 conform to our expectations. Age, holding
constant for the first and last years of commercial operation is everywhere negative. The
quadratic term proved to be statistically insignificant; the estimated effect is everywhere
negative. Operating efficiency declines at nearly one-half of a percent per year. Prior to
commercial operation, plants operate at load factors 35 percent lower than after they
begin commercial production. Also, load factors are 20 percent lower in the first year and
25 percent lower in the last year of commercial operation. Finally, shutdown plants were
7 percent less efficient in each year of the their commercial lives compared to plants that
are still running.
The effects associated with these variables hold constant general effects
associated with nuclear power plant operation in each year in each country. In other
words, we estimate a country and time specific factor of performance which essentially
averages operating efficiency for each country for each year. We assume that the startup,
shutdown, age, and construction experiences are common across countries and time.
However, we imagine that on top of these, there are country-specific factors. We allow
these to vary by time as well. In essence what we have is a yearly average of operating
efficiencies across all of the nuclear power plants in service in each country.
3.2.
Country Rankings
Table 8 shows the estimated country effects derived from the regression analysis
presented in Table 7. These estimated effects are load factors for the average reactor in
each country over its commercial life under the assumption of best-case construction and
30
There are several reasons why this might be true. Regulatory delay could have resulted in
mandated changes in design. Construction delays could have resulted from design flaws. Economic delays
could lead to redesign difficulties. Our estimate is the average over all of these.
17
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
startup. They represent the relative operating efficiencies across countries. Table 8 shows
the effects averaged over the years 1996 through 2002.31
Table 8. Average Load Factor Experience by Country
Country
Load Factor
Country
Armenia
0.53
Japan
Argentina
0.81
Korea, South
Belgium
0.90
Kazakhstan
Bulgaria
0.54
Lithuania
Brazil
0.69
Mexico
Canada
0.77
Netherlands
Switzerland
0.91
Pakistan
China
0.72
Romania
Czech Republic
0.84
Russia
Germany
0.87
Sweden
Spain
0.90
Slovenia
Finland
0.95
Slovak Republic
France
0.75
Taiwan
United Kingdom
0.80
Ukraine
Hungary
0.89
United States
India
0.62
South Africa
Load Factor
0.79
0.85
0.26
0.44
0.81
0.85
0.34
0.81
0.62
0.77
0.85
0.72
0.84
0.69
0.89
0.80
Notes: Load factor is based on regression analysis in Table 7, and averaged for years 1996 on. Estimates are
the average experience in each country under the assumption of a 30 commercial life and best-case
construction record.
Again we see a wide range of operating efficiencies. For the most part, the
relative rankings between countries change little from the raw data, though there are a
few interesting differences. Brazil, which seemed pathetic in the raw data, looks a little
better in these estimates. But, India and the eastern European countries including Russia
still bring up the rear while Finland, Belgium, and Switzerland are at the top.
An important question involving these operating efficiencies is, how much can be
attributed to the design and operation of the plants themselves and how much is due to
the overall electricity system of the country itself? This question is important in an
analysis of the cost of nuclear power because design and operational inadequacies are
potentially resolved if and when new plants are constructed, whereas inefficiencies
systemic to the countrywide electricity system are much less likely to be remedied. We
approach this question by examining the operating efficiency of the electricity system in
each country.
31
Data for most countries goes only through 2001.
18
Maloney & Diaconu
3.3.
LLNL Project: Sub-Contract B529375
Line Losses
The best measure of the operating efficiency of a country’s electricity system
available to us is line losses. Line losses represent electricity that is generated but lost in
the movement of power from the generator to the ultimate consumer. It is energy that is
dissipated in the transmission and distribution system.
More electric power is lost when power moves across low voltage lines than it
does at high voltage. For instance, in the United States, most of the large generation units
and especially the nuclear power plants are tied into the electrical system or grid through
500,000 volt lines. This power moves throughout the system to substations where the
voltage is reduced and reduced until it reaches households at 220 volts. Obviously, the
farther power moves along higher voltage lines in its path from generator to home, the
lower will be the line losses.
Country by country data on electricity generation, consumption, and line losses
from transmission and distribution were obtained from the World Bank.32 The data
include generation by type of fuel as well as total generation. The data are annual from
1992 through 2000. The average line loss percentage is shown in Table 9 for the
countries for which data are available.
The line loss data seem reasonable in the large. Developed countries typically
have lower line losses than underdeveloped ones. Even so, there are some outliers. One
way of examining these data is to relate the line losses to other characteristics of the
electrical system. We do this by estimating a regression of line losses on various
characteristics of the electricity system. Specifically, we regress line losses on the
percentage of exports, electricity consumption per capita, electricity consumption divided
by the area of the country, and area itself.
Exports are expected to be associated with lower line losses because exports are
almost always accomplished over high voltage transmission lines. Also, it seems
reasonable to believe that the transmission system will be of higher quality in countries
that have more exports. Across our sample the average export percentage is 5 with a
standard deviation of 10.33 Electricity consumption per capita and per square mile are
likely associated with lower line losses simply because they are indicators of more
intense electricity use.
32
The Energy Information Administration of the U.S. Department of Energy reports international
data from which line losses can be calculated. However, for more than half of the observations, the line loss
percentage is exactly 7. EIA cautions on the accuracy of these data; we second.
33
Export percentage is kilowatt hours of electricity exports divided by generation plus imports.
These data come from EIA-DOE.
19
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
Table 9. Electricity Line Losses in Transmission and Distribution by Country
Line
Line
Loss
Loss
Country
Percent
Country
Percent
Albania
48.7
Germany
4.0
Algeria
17.2
Ghana
0.8
Angola
21.8
Greece
7.5
Argentina
17.0
Guatemala
16.3
Armenia
32.2
Haiti
47.4
Australia
6.8
Honduras
23.9
Austria
6.3
Hong Kong, China
13.1
Azerbaijan
15.6
Hungary
12.9
Bahrain
4.7
Iceland
6.6
Bangladesh
17.7
India
20.7
Belarus
13.7
Indonesia
12.2
Belgium
5.0
Iran, Islamic Rep.
12.5
Benin
71.5
Ireland
8.7
Bolivia
21.1
Israel
4.2
Bosnia & Herzegovina
18.2
Italy
7.4
Brazil
16.7
Jamaica
10.4
Brunei
2.6
Japan
3.6
Bulgaria
14.0
Jordan
9.8
Cameroon
18.8
Kazakhstan
14.9
Canada
7.2
Kenya
18.4
Chile
9.0
Korea, Rep.
4.7
China
7.0
Kyrgyz Republic
24.4
Colombia
22.2
Latvia
29.4
Congo, Dem. Rep.
4.2
Lebanon
15.1
Congo, Rep.
38.6
Lithuania
14.0
Costa Rica
7.6
Luxembourg
25.4
Croatia
18.3
Malaysia
8.4
Cuba
18.0
Malta
10.8
Cyprus
5.6
Mexico
14.1
Czech Republic
7.6
Moldova
25.0
Denmark
5.7
Morocco
4.3
Dominican Republic
27.1
Mozambique
29.9
Ecuador
22.9
Myanmar
34.2
Egypt, Arab Rep.
12.1
Nepal
21.4
El Salvador
13.9
Netherlands
4.3
Estonia
16.7
Netherlands Antilles
12.4
Ethiopia
10.0
New Zealand
11.3
Finland
4.0
Nicaragua
26.4
France
5.9
Nigeria
32.4
Gabon
10.3
Norway
7.3
Georgia
19.8
Oman
14.5
Country
Pakistan
Panama
Paraguay
Peru
Philippines
Poland
Portugal
Qatar
Romania
Russian Federation
Saudi Arabia
Senegal
Singapore
Slovak Republic
Slovenia
South Africa
Spain
Sri Lanka
Sudan
Sweden
Switzerland
Syrian Arab Republic
Taiwan, China
Tajikistan
Tanzania
Thailand
Trinidad And Tobago
Tunisia
Turkey
Turkmenistan
Ukraine
United Arab Emirates
United Kingdom
United States
Uruguay
Uzbekistan
Venezuela
Vietnam
Yemen, Rep.
Zambia
Zimbabwe
Line
Loss
Percent
24.4
21.2
1.8
16
15.5
11.8
9.9
6.6
11.3
10.2
8.5
13.6
4.4
7.2
5.4
7.6
9.2
18.2
25.5
7
5.9
26.3
5.2
12.3
18.5
8.7
9.1
10.3
16.6
11.2
13.6
9
8.3
6.7
17
8.9
21
19.1
22.1
2.8
11.6
Notes: Data from the World Bank. Means of annual observations, 1992 through 2001.
The results of this regression are shown in Table 10. The equation explains 30
percent of the variation in line losses across time and places. All variable behave as
expected. The coefficient on the percentage of exports can be interpreted to say that a 10
percentage point increase in exports decreases line losses by 1.2 percentage points, so the
effect is not huge. The effect of electricity consumption per capita is statistically
significant, but also quite modest in its impact. The coefficient says that a 10 percent
increase in consumption per capita decreases line losses by .3 percent. Finally, electricity
consumption per square mile is trivial. The main point of this regression is simply to
20
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
show that line losses are systematically related to the electricity system, which gives us
some confidence about the quality of the data.
Table 10. Regression of the Percentage Line Losses on Country-Wide Electricity
System Characteristics
Independent Variables:
Coefficient
t-stat
Percentage of Exports
-0.12
-5.19
Electricity Consumption Per Capita*
-0.03
-11.31
Electricity Consumption Per Square Mile*
-3.75E-3
-1.83
Intercept
.41
27.93
R-squared
.31
No. of Observations
1074
Notes: (*) denotes logs.
Our main goal is to relate the quality of the electricity system to load factors for
nuclear generators. The results of this analysis are presented in Table 11. Here we regress
our estimated load factors on the line loss percentage. Several specifications are given.
Table 11. Regressions of the Estimated Load Factors for Nuclear Generators on
Country-Wide Line Losses
Coefficient / (t-stat)
Independent Variables:
(a)
(b)
(c)
Line Loss Percentage
-1.69
(-11.49)
Percentage of Exports
R-squared
No. of Observations
-1.66
(-5.07)
-0.18
(-1.83)
Electricity Consumption Per Capita*
Intercept
-1.40
(-6.20)
0.03
(1.85)
0.90
(52.29)
.33
275
0.66
(4.83)
.34
273
0.90
(22.36)
.46
32
Notes: (*) denotes logs. Specification (c) is country averages over the years for which line loss data are available.
Specifications (a) and (b) are based on pooled time series and cross sectional observations. The number of
observations differs because of data availability on electricity exports.
The results shown in Table 11 demonstrate with a reasonable degree of precision
that load factors at nuclear power plants are significantly related to the overall quality of
the electricity system as measured by the percentage line losses. The estimated
coefficient says that a 1 percentage point increase in line losses is associated with a 1.5
percentage point decrease in load factor. The estimated coefficient varies by a statistically
insignificant amount based on specification. The specification based on the average of the
annual observations for each country assures us that the statistical significance of the
effect is not spuriously inflated by autocorrelation. Inclusion of the other variables in the
regression does not affect the result, nor are they significant predictors of nuclear
generator load factors.
21
Maloney & Diaconu
3.4.
LLNL Project: Sub-Contract B529375
Application of the Results
Given our estimated relation between nuclear power plant operating efficiency
and the overall efficiency of the electric system, it remains only to predict the values for
countries of interest. Table 12 shows the actual and predicted values of load factor for all
countries currently operating nuclear power plants and for three countries that are
considering nuclear generation: Egypt, Iran, and Turkey.
In comparing the predicted values for countries with nuclear power plants in
operation, notice that there is some variation between the actual and predicted values.
The model does not explain all variation (as shown by the R-squared statistic), but it is
correct on average. It is interesting to note that two countries of interest, China and India,
both have predicted values that are very similar to the actual ones. For India, the
predicted and observed operating efficiencies are below par. The actual and predicted
load factor for China is only slightly below the benchmark level.
For the three countries that are considering nuclear power, our forecast of
operation efficiencies are all also below par. By our estimates, both Egypt and Iran can
both expect nuclear power to be around 11 percent more expensive than that based on
even a conservative scenario. Turkey can expect nuclear power to be 20 percent more
expensive.
Table 12. Predicted Nuclear Power Plant Load Factors Based on
Country-wide Line Losses
Load Factor
Line Loss
Predicted Load
Country
Experience
Percent
Factor
Argentina
0.85
0.17
0.61
Armenia
0.53
0.32
0.36
Belgium
0.87
0.05
0.81
Brazil
0.47
0.17
0.62
Bulgaria
0.53
0.14
0.66
Canada
0.74
0.07
0.78
0.77
0.07
China
0.78
Czech Republic
0.82
0.08
0.77
Finland
0.93
0.04
0.83
France
0.71
0.06
0.80
Germany
0.84
0.04
0.83
Hungary
0.89
0.13
0.68
0.53
0.21
India
0.55
22
Maloney & Diaconu
Japan
Kazakhstan
Korea, South
Lithuania
Mexico
Netherlands
Pakistan
Romania
Russia
Slovakia
Slovenia
South Africa
Spain
Sweden
Switzerland
Taiwan
Ukraine
United Kingdom
United States
Selected
Countries
Egypt
Iran
Turkey
LLNL Project: Sub-Contract B529375
0.77
0.35
0.84
0.42
0.79
0.85
0.37
0.79
0.61
0.74
0.80
0.72
0.86
0.75
0.90
0.81
0.67
0.80
0.83
0.04
0.15
0.05
0.14
0.14
0.04
0.24
0.11
0.10
0.07
0.05
0.08
0.09
0.07
0.06
0.05
0.14
0.08
0.07
0.84
0.65
0.82
0.66
0.66
0.82
0.49
0.71
0.73
0.78
0.81
0.77
0.74
0.78
0.80
0.81
0.67
0.76
0.78
0.12
0.13
0.17
0.69
0.69
0.62
Notes: Load factor experience is our measure of the operational performance on average in
each country for nuclear power plants over thirty years of commercial life assuming a bestcase construction record. These numbers differ slightly from the earlier table because they are
averaged over the years for which line loss data is available.
4. Summary & Conclusions
This document reports our investigation of two aspects of the business
environment in which Minatom operates. As we stated in the introduction, we see two
major questions in the nuclear construction business initiatives of Minatom: (i) Is it
possible that Minatom is making money on nuclear power plant construction given the
price that it is charging for this work? (ii) Are countries investing in nuclear power
making economically efficient decisions? To help answer these question, we have
examined the various estimates of construction cost for nuclear power plants, and we
have looked at country-specific factors affecting the operating efficiency of nuclear
power plants.
The basic conclusions reached in this report are the following:
1) There is a great deal of variance in the forecast capital cost of nuclear power plants.
This is true not just for Russian construction but also for costs of other reactors such
as the Canadian and French models. Nonetheless, it is clear that forecast cost of
Russian units installed internationally is typically lower than the rest of the
competition. This suggests that Russia is consistently low-bidding. The difference
between the price Russia is charging and the price charged by competitors in the
23
Maloney & Diaconu
LLNL Project: Sub-Contract B529375
international construction industry runs somewhere between 15 and 25 percent. This
is roughly equivalent to the technology component in the cost structure plus minimal
accounting for contingencies and interest charges during construction.
2) The capital cost of electricity per kWh is inversely proportional to load factor. Hence,
a ten percent decrease in load factor translates approximately into a ten percent
increase in the full price of electricity. The historical, worldwide average load factor
for nuclear power plants is around 70 percent; this compares to assumed load factors
of between 80 and 90 percent used in nuclear power plant cost projections. We think
that this is an important factor in assessing the economic efficiency of new nuclear
generation projects, such as those in Iran, India, China, and proposed projects in
Egypt and Turkey.
We search for factors that are systematically linked to load factors at nuclear power
plants. The objective is to link the factors associated with the operating efficiency of
the electricity system of a country to load factor so that we can more accurately
predict the expected load factor for a proposed nuclear power plant. We find that
countrywide nuclear power plant load factors are systematically linked to the line
losses experienced within each country. Our estimates say that a 1 percentage point
increase in line losses is associated with a 1.5 percentage point decrease in load
factor.
Based on this, we are able to forecast load factors for nuclear power plants even for
countries that are not yet employing nuclear generation technology. For three such
countries of interest, Egypt, Iran, and Turkey, we forecast nuclear power plant load
factors much lower than the baseline assumptions. This very likely makes nuclear
power an uneconomic source of generation for these countries.
24