Issues
In Focus.
Nuclear Power Plant Construction Costs
With the improved performance of the 104 operating
U.S. nuclear power plants, increases in fossil fuel prices, and
concerns about global warming, interest in building new nuclear
power plants has increased. Because no nuclear plants have been
ordered in the United States in nearly three decades, the costs
of a new plant are uncertain. To assess the economics of building
new nuclear power plants, EIA conducted a series of workshops and
seminars focusing on key factors that affect the economics of nuclear
powerprimarily, the cost of building power plants and the
financial risks of constructing and operating them.
History of Nuclear Power Construction Costs
As was typically the case with fossil-fuel-fired
power plants, many of the first-generation U.S. reactors were constructed
on a fixed price, turnkey basis. Under this type of contractual
arrangement, the vendor assumed all the risk associated with cost
overruns and scheduling delays. In total, about 12 units were ordered
on a turnkey basis in the early to mid-1960s. Although the costs
of the reactors were never made public, one study estimated that
the vendors lost more than $1 billion [71]. As a result,
they eventually stopped offering turnkey contracts to build nuclear
power plants and instead went to cost-based contracts.
Factors affecting the costs of non-turnkey U.S. reactors
have been the subject of a number of analyses. An EIA analysis found
that realized real overnight costs grew from about $1,500 per kilowatt
for units beginning construction in the 1960s to about $4,000 per
kilowatt for units beginning construction in the early to mid-1970s
(all costs in 2002 dollars, except where noted). Lead times also
increased, from about 8 years to more than 10 years. Much of the
growth in overnight costs and lead times was unforeseen by those
preparing the estimates, and overruns in real overnight costs and
lead times ranged from 70 to 250 percent [72].
Because of severe data limitations and the inherent
difficulty in measuring regulatory impacts, there is only qualitative
agreement that the following factors caused the growth in nuclear
plant costs and lead times [73]:
- Increased regulatory requirements that caused design changes
(backfits) for plants under construction
- Problems in managing mega projects
- Misestimation of cost savings (economies of scale) for larger
plants
- Misestimation of the need for the capacity.
Historically, the deployment of nuclear plants abroad
lagged behind that in the United States. Thus, there was a tendency
for utilities in Europe and Asia to learn from the U.S. experience.
Now, just the opposite is occurringthe next generation of
U.S. nuclear power plants will benefit from foreign learning. Accordingly,
EIAs present cost estimates used realized costs of nuclear
power plants in Asia as a starting point.
Building New Nuclear Plants in the United States
One of the major uncertainties in building new nuclear
power plants involves the regulatory and licensing process. Regulatory
actions were one of the factors that contributed to the cost growth
in the 1970s and 1980s, and as a result there were significant efforts
to reform the process. In the late 1980s, the U.S. Nuclear Regulatory
Commission (NRC) modified backfit regulations to make it more difficult
to order changes in a plants design during construction. Additionally,
with the passage of the Energy Policy Act of 1992, the licensing
process was also changed substantially. Before 1992, a utility needed
one license to begin construction and another to begin commercial
operation. Public hearings were a prerequisite for both licenses,
and in some cases they proved to be very contentious. Now, as long
as a firm follows all the agreed-upon procedures, tests, and inspections,
separate hearings are not required. The 1992 legislation also allowed
for the pre-approval of various designs; as a result, many technical
engineering issues can be settled before the licensing process begins.
Beginning in the mid-1990s, the nuclear industry
began to design new Generation III (or III+) reactors. In general,
the new designs represent incremental improvements over the current
generation of light-water reactors. They are simpler and include
more passive safety features. As discussed below, these
design changes have cost implications.
The vendors of two Generation III reactorsthe
Advanced Boiling Water Reactor (ABWR) and an Advanced Pressurized
Water Reactor (the AP1000) have provided estimates of construction
costs. GEs estimate for the ABWR ranges from $1,400 to $1,600
per kilowatt (2000 dollars) for a large, single-unit plant (1,350
megawatts or more). British Nuclear Fuels Limited (BNFL), the manufacture
of the AP1000, has estimated that construction costs for the first
two-unit 1,100-megawatt reactors will range from $1,210 to $1,365
per kilowatt (2000 dollars). GEs estimate assumes that the
government would pay for 50 percent of the first-of-a-kind engineering
costs, and BNFLs estimate assumes that the government (or
someone other than the purchaser of the plant) would pay for all
the first-of-a-kind costs. BNFL also assumes that, because of learning,
a third two-unit plant could be built for about $1,040 per kilowatt
(2000 dollars) [74].
A state-owned Canadian firm, Atomic Energy Canada
Limited (AECL), has also stated its intention to market an advanced
CANDU reactor, the ACR-700, in the United States. The ACR-700, a
design that uses heavy water to moderate the reaction, is substantially
different from the AP1000 and ABWR [75]. One major advantage
of CANDU reactors, which have been built worldwide [76],
is the ability to refuel the unit while it is operating. Light-water
reactors must be taken out of service before they can be refueled.
On the other hand, the use of heavy water raises nuclear proliferation
issues. The total cost of building third of a kind twin-unit
plants has been estimated by AECL at about $1,100 to $1,200 per
kilowatt.
All the above estimates are much lower than the capital
costs that have been realized in the past for nuclear power plants
built in the United States and abroad [77]. As noted above,
the average construction cost of U.S. units that entered commercial
operation in the 1980s was about $4,000 per kilowatt. On average,
light-water and CANDU reactors have been built in the Far East and
elsewhere abroad at costs that are in the low $2,000s per kilowatt.
The AP1000 has never been built anywhere in the world. If the vendors
are able to achieve their projected costs, their plants are likely
to be competitive with other generating options. The key question
is whether cost reductions of the magnitude projected by the vendors
are achievable.
There is reason to believe that new reactors will
be less costly to build than those currently in operation in the
United States. Over the past 30 years, there have been technological
advances in construction techniques that would reduce costs. In
addition, the simplified, standardized, and pre-approved designs
clearly result in cost savings. The newer plants have fewer components
and therefore would be less costly. At least in the United States,
only a few previously built plants were based on standardized designs,
and in most cases construction began before the unit was totally
designed. The construction of customized units, with the design
work being done during the plants construction, is clearly
expensive. Because the designs of advanced reactors are (or will
be) pre-approved by the NRC, much of the design work will be done
before their construction begins, and this will lower costs. Regulatory
changes will also lower regulatory costs and risk.
Although it is reasonable to expect lower construction
costs for the new reactors, EIA and other organizations have questioned
the size of the cost reductions [78]. This is particularly
true of the vendors estimates relative to recently realized
costs in Asia.
All the cost estimates from nuclear vendors assume
savings from building large multi-unit plants. The estimates for
the AP1000 and CANDU reactors assume two unit sites, and those for
the ABWR deal with a 1,350- to 1,500-megawatt reactor. As discussed
below, the size of these projects has financial implications that
cannot be overlooked. Moreover, there is some evidence that cost
overruns for earlier U.S. reactors resulted from misestimation of
the savings from building large or multi-unit plants.
There are four major parties (and numerous secondary
ones) involved in the construction of a nuclear power plant: a firm
that manages the construction of the plant, a firm that supplies
engineering and architectural support, a firm that supplies the
reactor or Nuclear Steam Supply System, and the firm that purchases
the unit. All incur costs, and it is important that all their costs
be included in the estimate. It is possible that some reported estimates
might deal only with the costs to two or three of the parties; in
such cases, the estimates would not be inclusive.
Results of EIA-Sponsored Workshops and Seminars
and Derivation of EIA Estimates
In addition to sponsoring several workshops and seminars
on the subject of nuclear construction costs, EIA also commissioned
a series of reviews of the vendor estimates. All the reviewers generally
found that the estimates included the costs to the four parties
involved with the construction of a nuclear power plant, but they
also found that the estimates were not sufficiently detailed to
permit verification of their accuracy. Indeed, the only way to verify
the estimates would be to reproduce theman effort that is
prohibitively expensive.
EIAs reviewers were forced to use their subjective
judgment, and there were differing opinions about the estimates.
The reviewers and workshop participants from the nuclear industry
think that the cost reductions are achievable, making arguments
similar to the ones presented above. One reviewer who is an outside
observer of the industry, one workshop participant who is a financial
analyst, and some outside researchers were more skeptical. For example,
in a recent study from the Massachusetts Institute of Technology
(MIT), researchers used $2,000 per kilowatt as a base case
and employed a 25-percent cost reduction as unproven but plausible.
The procedure used to derive nuclear construction
cost estimates for AEO2004 is as follows. For non-nuclear
technologies, EIA uses cost estimates consistent with realized outcomes
for the construction of new generating capacity in the United States.
However, because no reactors have been built recently in the United
States, EIAs cost estimates are based on foreign cost data.
There are two marketable Generation III light-water reactors currently
in operation, and another four are under construction in Asia [79].
Thus, the starting point for an estimate of building the next
new U.S. advanced nuclear power plant was the realized cost of the
two operating light-water nuclear units in Asia. In AEO2004,
$2,083 per kilowatt (inclusive of all contingencies) is used as
the realized cost for these two reactors [80].
The four units that are under construction in Asia
will be completed over the next 5 years. The first new U.S. plant
could not become operational until 2012 at the earliest. Thus, the
construction of the first U.S. plant will benefit from experience
gained in the construction of the four units in Asia.
For all advanced technologies that are in the early
stages of commercialization, EIA assumes that, because of learning,
U.S. capital costs will fall by 5 percent for each of the first
three doublings of newly built capacity. The same learning factor
is applied to the costs of the four advanced light-water reactors
under construction in Asia. Thus, the cost reduction from learning
in building four additional reactors (roughly 1.5 doublings of capacity)
is about 8.5 percent. As a result, the assumed realized cost, inclusive
of contingencies, of the sixth advanced light-water reactor in Asia
when it is completed is $1,928. This is the estimate used in the
projections [81].
As new U.S. nuclear plants are built, because of learning,
EIA assumes that costs will continue to fall. For example, if 10
new units were constructed in the United States, costs would continue
to fall to about $1,719 per kilowatt (inclusive of all contingencies)
as a result of learning. Even if no nuclear plants were built in
the United States, EIA assumes that costs would fall to about $1,752
per kilowatt by 2019. As shown in Figure 36, the AEO2004
cost estimates are below realized costs for older U.S. plants and
plants recently built abroad.
The vendors estimates of construction lead
times are generally about 36 to 48 months from the date of the first
concrete pour to the date of initial system testing (or fuel loading).
This definition of lead time is often used, because most of the
funds are expended over that period. To compute interest costs,
EIA uses a slightly different definition of lead timesnamely,
the time between the commencement of the licensing process to the
date of commercial operation. The licensing process will take 12
to 24 months, and there will be an additional 6 months between fuel
loading and commercial operation. Thus, EIA assumes a 6-year lead
time.
In one of EIAs workshops, the issue of the
time and cost for preparing a license application and the expenses
incurred in obtaining the license were discussed. Some within the
industry think an additional 4 years would be needed to prepare
the application and license the first few plants, resulting in a
10-year total lead time. A small cost premium (up to 5 percent)
is added by EIA to the cost of just the first four units built.
This is called the technological optimism factor. Because
this factor gradually goes to zero as new nuclear plants are constructed,
there will be an additional reduction in costs over and above the
learning effects. This cost reduction, in part, captures the reduction
in expenses associated with the 4-year reduction in lead times as
a result of improvements in the licensing process.
Summary of the Projections
Over the past few years, most economic analyses of
nuclear power have tended to compare the cost of generating electricity
from nuclear technology with the cost of producing power from a
combined-cycle natural-gas-fired power plant. As long as natural
gas prices remain in the range of $2 to $3 per thousand cubic feet,
the cost of building and operating a new gas-fired plant will be
much less than the cost of a new coal-fired plant. Therefore, the
assumption has been that nuclear power would compete with combined-cycle
gas plants. With natural gas prices rising, however, new coal-fired
power plants and, to some extent, renewable energy are becoming
competitive with new natural gas units in many parts of the United
States.
The AEO2004 reference case assumes that nuclear
power plant construction costs will fall from $1,928 per kilowatt
to $1,752 in 2019. On that basis, no new nuclear power plants would
be built before 2025 in the reference case. In two advanced nuclear
cases, vendor estimates for the AP1000 and ACR-700 reactors are
used. In both advanced cases, the current level of nuclear capital
costs is assumed to be lower than in the reference case, and cost
reductions are assumed to be greater than in the reference case.
Specifically, one advanced casethe vendor estimate caseis
based on an average of the AP1000 and ACR-700 reactor first-of-a-kind
and nth-of-a-kind costs [82]. In this case, costs
would fall from $1,555 per kilowatt in 2004 to $1,149 in 2019. The
second advanced nuclear casethe AP1000 caseuses just
the vendor cost estimates for the AP1000. In this case, costs would
fall from $1,580 per kilowatt to $1,081 in 2019.
In the AP1000 case, where costs fall to about $1,081
per kilowatt in 2019, EIA projects that about 26 gigawatts of new
nuclear power plant capacity would be constructed and become operational
by 2025. The 26 gigawatts of new nuclear power plant capacity would
displace 19 gigawatts of coal-fired capacity and 7 gigawatts of
mainly fossil-fuel-fired capacity. In the average cost case, where
costs fall to $1,149 per kilowatt in 2019, 12.8 gigawatts of new
nuclear power capacity would be built and become operational by
2025, displacing about 9.4 gigawatts of coal-fired capacity.
If the projections were extended beyond 2025, or
if the cost reductions occurred more rapidly than assumed in the
two advanced nuclear cases, the projected amount of new nuclear
capacity would be much greater. The total assumed capital cost of
a pulverized coal plant in 2005 is $1,170 per kilowattabout
10 percent higher than the vendors estimate of the AP1000
costs [83]. Coal and nuclear fuel costs are 10 mills and
4 mills per kilowatthour, respectively. Historically, non-fuel operating
and maintenance costs are roughly the same for the two technologies.
Given a nuclear capital cost estimate of $1,081 per kilowatt, both
the capital and operating costs would therefore be less for nuclear
than for coal-fired power plants. If the $1,081 per kilowatt estimate
could be realized, it is possible that nuclear power could eventually
be used to satisfy virtually all the baseload demand in the United
States in future years.
The Issue of Risk
Another issue that received considerable attention
in the EIA workshops was the financial risk in constructing and
operating any power plant. There are risks associated with the use
of natural gas, coal, and nuclear power. Natural-gas-fired power
plants can be built in a few years and are relatively inexpensive,
and thus there is little risk in their construction; however, because
natural gas prices are volatile, there are risks involved with the
operation of gas-fired power plants. Indeed, a number of the workshop
participants noted that nuclear power can be used to hedge fuel
price risks associated with gas plants.
Environmental factors aside, coal prices are relatively
stable, and thus the fuel price risks associated with coal-fired
power plants are small. Environmental regulations could change,
however, especially with respect to global warming, with major impacts
on the economics of operating coal plants. Thus, there are regulatory
risks associated with the operation of coal-fired power plants.
One workshop participant noted that firms have been able to finance
the construction of coal-fired plants because of a perception that
changes in environmental regulations will not occur for another
10 to 15 years, and by then the loans will have been repaid.
There are also regulatory risks involved with the
construction and operation of nuclear power plants. According to
a number of workshop participants, the financial community clearly
has not completely discounted the cost overruns that occurred in
the 1970s and 1980s. Thus, all the participants agreed that the
nuclear industry must demonstrate that a nuclear power plant can
be built on time and on budget. Further, the new licensing process
has yet to be tested, and there is considerable uncertainty about
how it will work. In fact, all the participants agreed that some
type of support from a third party (the Federal Government) would
be needed before the first few plants could be built.
If nuclear power plants are built in a deregulated
environment, their ownerslike the owners of any power plantwill
be exposed to output price risk. Electricity prices might be lower
than anticipated, resulting in insufficient revenues to cover all
the operating costs, loan repayments, and returns to shareholders.
As a result of market deregulation, electricity is now a commodity,
and like any other commodity, in the short run electricity prices
are extremely volatile and subject to boom and bust
cycles. The events of the past few years suggest that if plants
become operational in the bust part of a cycle, the
result can be financial ruin.
Although all units are subject to output price risk,
nuclear power plants are affected differently because of their relatively
high capital costs and longer lead times. That is, because of nuclear
powers relatively high capital costs, relatively more capital
is at risk. Moreover, the uncertainty of any forecast
of electricity prices increases as the length of the forecast period
increases (a 6-year forecast is more uncertain than a 2-year forecast).
Because of nuclear powers relatively long lead times, electricity
prices must be anticipated over a relatively long period, leading
to more uncertainty.
All the workshop participants outside the nuclear
industry argued that stable and predictable revenues resulting from
long-term, fixed-price power purchase agreements or other financial
or regulatory instruments are crucial to the financing of a nuclear
power plant. Long-term (10 to 20 years) firm fixed price purchased
power contracts are, however, very difficult and expensive to obtain.
Moreover, as a recent EIA report noted, until some structural flaws
in electric power markets are corrected, the use of financial derivatives
to manage electricity price risk is limited [84]. Thus, at
least in the short run, it is not clear whether it will be possible
to obtain a stable stream of revenues from a nuclear (or other)
power plant.
The advanced nuclear cases summarized above and presented
in detail in the Market Trends section of this report
assume that institutional and financial arrangements can be used
to mitigate (or shift) output price risk at very little cost to
decisionmakers. A fixed-price purchased power contract is one possible
financial arrangement that would shift the risk to those holding
the contract. Another possible institutional arrangement would be
a consortium formed by a group of utilities and vendors to build
nuclear power plants. In such a case, the risks would be spread
among all the consortium members.
Notes and Sources
Released: January 2004
|