Professional Documents
Culture Documents
Energy Policy
journal homepage: www.elsevier.com/locate/enpol
Is it the end of the line for Light Water Reactor technology or can China and T
Russia save the day?
Steve Thomas
Emeritus Professor of Energy Policy, Public Services International Research Unit (PSIRU), Business School, University of Greenwich, 30 Park Row, London SE10 9LS, UK
Keywords: In the late 1990s, a new generation of reactor designs evolved from existing designs was touted as solving the
Light Water Reactors economic problems that led to the collapse of reactor ordering after the Chernobyl disaster. It was claimed these
Nuclear power designs would be cheap and easy to build because they would be simpler and use passive safety, modular
Russia construction and standardisation. The US and UK governments were convinced by this and launched reactor
China
construction programmes. However, 20 years on, the claims have proved false and the US and UK programmes
Nuclear renaissance
are in disarray. The last hope for the nuclear industry appears to be that Chinese and Russian reactor vendors,
with powerful support from their governments, will take over, providing reactors that are cheap but meet the
safety standards required in Europe and North America. However, these vendors and their designs are largely
unproven in open markets. There is also little evidence that their reactors will be cheap, there are concerns about
quality and safety culture and there are national security concerns that may deter customers. New technologies,
such as radical new ones, Generation IV, and Small Modular Reactors are unproven and, at best, a long way from
commercial deployment
https://doi.org/10.1016/j.enpol.2018.10.062
Received 20 June 2017; Received in revised form 6 October 2018; Accepted 28 October 2018
0301-4215/ © 2018 Elsevier Ltd. All rights reserved.
S. Thomas Energy Policy 125 (2019) 216–226
and Westinghouse, which allowed them to dominate the market rather of reactors was at least US$7000 per kW.10
than any superiority over the alternative reactor types.4 By the mid-70s
and before the 1978 Three Mile Island (TMI) accident, problems of cost 2.2. Safety
were serious and Bupp and Derian (1978) in their book ‘Light Water:
How the nuclear dream dissolved’ talked about the ‘extravagance of There was no specific claim on safety across all designs other than
prophesy’ particularly on costs, that had led to the collapse of reactor that they would be safer than existing ones. For example, Westinghouse
ordering in the USA.5 Reactor orders continued in Europe after TMI into claimed the core damage frequency for its AP1000 would be 5 × 10−7,
the 1980s, notably in France. However, even before the Chernobyl 1 per cent of the frequency for ‘current’ plants and less than one
disaster in 1986, there were strong signs very few orders would be thousandth of the requirements of the US Nuclear Regulatory
placed in the short- to medium-term. Committee (NRC).11 One of the ways to achieve this was use of ‘passive
Costs increased dramatically from when the first commercial orders safety’ under which, in an accident, there would be less reliance on
were placed.6 Lovering et al., (2016, 2017) argue that the picture of engineered safety systems such as emergency core cooling systems,
rapidly escalating real construction costs is not as universal as is often with the plant being kept in a safe condition by natural processes such
portrayed, bringing in data from countries such as South Korea and as convection.
India. However, Koomey et al. (2017) argue that the reliability of this For some designs there was also the addition of ‘core-catchers’. If
data is questionable and given that it includes data from technologies there was a core meltdown, the core would be ‘caught’ and prevented
other than LWRs and is often for plants completed long ago, its re- from entering the environment. Following the 9/11 attack in 2001,
levance to this analysis is questionable. The Three Mile Island and there was also a requirement to require containment buildings to be
Chernobyl accidents showed that serious nuclear accidents were a real able to survive an impact by a large aircraft.
not just a theoretical possibility. For the decade following Chernobyl
there were few orders. The cost of power from nuclear reactors was 2.3. Financeability
clearly far higher than power from coal or gas; financiers were reluctant
to lend money for nuclear construction because of its poor record of Even if new reactors were forecast to be cost competitive, financiers
being built to time and cost; and the opening of electricity to compe- would be reluctant to lend money for reactor construction because of
tition meant utilities were less able to recover costs of expensive options the poor record of reactors being built to time and cost. The claim was
from consumers. that reactors would be simpler so easier to build. Greater simplicity
The future for nuclear power looked bleak and, to counter this the could be achieved, it was argued, by re-evaluating the safety systems
nuclear industry began to promote a new generation of reactors evolved that had been progressively added to reactors in response to accidents,
from existing LWRs (see Table 1).7 The publicity was designed to show rationalising them to achieve a better level of safety but with a simpler
that these new designs addressed the issues that were behind the col- design. The use of passive safety was expected to help.
lapse of ordering: cost; safety; and financeability. In the USA and the Another element in the strategy to reduce construction risk was the
UK, the governments tried to reinforce this process by introducing use of modularisation. Reactor construction was largely an on-site ac-
systems of comprehensive reactor design review that would approve a tivity. It was argued that site work is difficult to manage and if much of
standardised design for construction at any site. the construction could be shifted to factories with on-site work re-
stricted to assembling modules, the risk of cost and time over-runs
2.1. Cost would be reduced. Construction time was predicted to be four years or
less.
In the late 1990s, the cheapest form of generation was generally gas-
fired plants, occasionally coal for plants near cheap deposits of coal. To 2.4. Design review
be competitive, nuclear had to match the generating cost of gas and this
meant nuclear construction costs could be no more than US$1000 per In 1992, the US NRC introduced ‘Design Certification’.12 A Design
kW of installed capacity.8 It was claimed Gen III+ reactors would cost Certificate would mean the design was approved to be built anywhere,
no more than this. Whether this forecast was based on a realistic esti- subject to local issues, for a period of 15 years. The rationale for this
mate of costs or whether it was adopted simply because that was the was that under the previous regime, the design would be given outline
cost needed is a moot point. However, in retrospect, this forecast is approval before construction, but the details of the design would be
scarcely less ‘extravagant’ than ‘power too cheap to meter’9 claimed reviewed for each project during construction. This was said to have
more than 40 years previously. By 2015, the expected construction cost caused delays in construction when resolving design issues led to a
slow-down in the construction process. The first designs to complete
this process were the Combustion Engineering System 80+ and the GE
ABWR in 1997 and the Westinghouse AP600 in 1999.13 These designs
received no orders, and their approval has expired. The Westinghouse
AP1000 started its review in 2002 and was given approval in 2006 but
4
See for example, Cowan, R. (1990). Nuclear Power Reactors: A Study in Westinghouse submitted a series of design changes and the final design
Technological Lock-in. The Journal of Economic History, 50(3), 541–567. doi:10. was given approval only in 2011. Four reactors of this design began
1017/S0022050700037153.
5
No reactor order not subsequently cancelled was placed in the USA from
10
1974 to 2013 and more than 100 reactor orders, some almost complete, were For example, the proposed Hinkley Point C EPR is expected to cost £10bn
cancelled. per reactor or about US$8500/kW. https://www.gov.uk/government/news/
6
The first commercial order, when a nuclear power plant was chosen because hinkley-point-c-to-power-six-million-uk-homes (Accessed June 14, 2017).
11
it was expected to be the cheapest source of power was the US order for the https://www.iaea.org/NuclearPower/Downloads/Technology/meetings/
Oyster Creek BWR in 1963. 2011-Jul-4–8-ANRT-WS/2_USA_UK_AP1000_Westinghouse_Pfister.pdf
7
A more radical set of new designs using coolants and moderators other than (Accessed April 3, 2017).
12
light water known as Generation IV was also talked about these are acknowl- https://www.nrc.gov/reactors/new-reactors/design-cert.html (Accessed
edged to be decades away from commercial deployment. April 4, 2017).
8 13
This is a so-called ‘overnight’ cost and excludes financial charges. For an account of the designs that have received a Design Certificate see
9
This phrase was coined by Lewis Strauss for the National Association of https://www.nrc.gov/reading-rm/doc-collections/fact-sheets/new-nuc-plant-
Science Writers, September 16, 1954. des-bg.html (Accessed June 13, 2017).
217
S. Thomas Energy Policy 125 (2019) 216–226
Table 1
Generation III+ designs.
Design Vendor Passive safety Core-catcher Modular construction NRC approval ONR approval Under construction In service
construction in the USA in 2013/14. The ESBWR started the process in 3. The designs and experience with them
2005 and was given design certification in 2014. The Areva EPR began
its review in 2007 but in 2015, Areva applied to suspend the process There are four designs claimed to be Gen III+ that are under con-
because of the absence of US customers. The KEPCO APR1400 began its struction but by January 2018, only one reactor was in service: the
review in 2014. Toshiba and GE-Hitachi separately applied for Design Westinghouse AP1000 (eight under construction), the Areva EPR (four
Certification renewal for their versions of the ABWR (see below) in under construction), the Rosatom AES-2006 (one in service and five
2010 but Toshiba withdrew the request in 2016 with little progress under construction) and China's Hualong One (four under construction)
having been made. (see Table 1).18 The ABWR design supplied by GE, Toshiba and Hitachi
The UK introduced a similar design review process, Generic Design has been built in a 1980s version (four in service and four under con-
Assessment (GDA), in 2007.14 The EPR began the process in 2007 and struction). The design was updated in the 1990s for the US market and
was given a Design Acceptance Confirmation (DAC) in 2012.15 The has been updated for the UK market but not expected to be ordered
Westinghouse AP1000 began its review in 2007 but in 2011, Wes- before about 2020. The KEPCO APR1400 (one in service and seven
tinghouse paused it because of lack of customers, restarting it in 2014 under construction) does not meet Gen III+ standards but is being
with completion in 2017.16 The ESBWR started the review process in upgraded and is under review by the NRC. The GE ESBWR has no
2007 but was withdrawn a year later. The Hitachi-GE ABWR began its realistic order prospects and is not considered in detail.
review in 2013 and completed it in December 2017.17 The CGN Hua- The CAP1400 is a design developed in China by SNPTC based on the
long One started its review in 2017. AP1000 but scaled up to about 1400 MW. Construction had not started
A detailed critique of the generic review process is outside the scope on reactors of this design by January 2018. In 2010, Rosatom an-
of this article but there are serious issues. In some cases, design features nounced a new design, WWER-TOI that would be cheaper and quicker
are not specified in detail and are only resolved later when there is a to build than the AES-2006 with the expectation that orders for this
potential customer. This clearly means vendors minimise the money design would be placed from 2011 onwards. However, by 2018, no
they spend filling out the detail of the design until they have a cus- orders for this design had been placed. Other designs exist but have
tomer. For some elements of the design where technology is moving little prospect of winning orders in the short-term.
rapidly, for example, IT systems, it would make no sense to freeze the The lack of a precise definition of the characteristics of reactors that
design. As a result, only a small part of the design of the version of the can be categorised as Gen III+ means there is a significant diversity
AP1000 that received regulatory approval in 2006 was specified in full between the designs claimed to be Gen III+. The World Nuclear
detail and the design amendments were necessary only when US cus- Association stated that: ‘Generation III (and III+) are the advanced
tomers emerged. In practice the difference between generic review and reactors discussed in this paper, though the distinction from Generation
review on a site by site basis, as is still the case in jurisdictions other II is arbitrary.19 The designs can be divided into radical new ones and
than the UK and the USA, may be much less than at first sight. those developed from existing designs. The former category, which
Safety regulation is under national jurisdiction and regulators from includes AP1000 and ESBWR rely on passive safety and modularisation.
different countries impose different design requirements. For example, The latter includes EPR, AES-2006, ABWR and APR1400 and given that
in 2009, the French, Finnish and UK regulators jointly expressed con- additions will have been made to their predecessors such as core-
cerns about redundancy in the Instrumentation & Control (I&C) systems catchers, it seems implausible that these designs could be simpler than
for the EPR. However, the solution to this problem was different for their predecessors.
each country and for the UK, the design will only be specified at the
construction phase. While many countries will look to the analysis of 3.1. AP1000
experienced, and open regulatory regimes such as the USA, they may
still impose their own design requirements, so the standard design will The Westinghouse reactor division was sold by British Nuclear Fuels
tend to apply to only one country. The generic review system has been to Toshiba in 2007. The Advanced Passive 1000 (AP1000) was devel-
little tested in practice with only one design, the AP1000 in the USA oped from the AP600 which received US Design Certification in 1999
having started construction in the country it was certified. Whether a but was not marketed. The AP1000 was submitted for review in 2002
design certified more than a decade previously would still be close when the process was forecast by Westinghouse to be short because the
enough to the state of the art to be credible is a moot point. AP1000 represented only a scaled up AP600. Design Certification was
given in 2006 but design modifications were submitted soon after and it
was not until 2011 that final Certification was given. By then con-
14
http://www.onr.org.uk/new-reactors/assessment.htm (Accessed April 4, struction had already started on two pairs of AP1000s in China (see
2017).
15
http://www.onr.org.uk/new-reactors/reports/step-four/close-out/
18
epr70475n.pdf (Accessed June 13, 2017). A review of Gen III+ designs https://aris.iaea.org/Publications/IAEA_
16
http://news.onr.org.uk/2017/03/design-acceptance-for-the-ap1000- WRC_Booklet.pdf Accessed April 4, 2017.
19
reactor/ (Accessed June 13, 2017). http://www.world-nuclear.org/information-library/nuclear-fuel-cycle/
17
http://news.onr.org.uk/2017/12/regulators-approve-new-nuclear-power- nuclear-power-reactors/advanced-nuclear-power-reactors.aspx (Accessed April
station-design/ (Accessed January 19, 2018). 4, 2017).
218
S. Thomas Energy Policy 125 (2019) 216–226
Table 2 Table 3
Construction record of the AP1000. Construction record of the EPR.
Source: Various. Source: Various.
Site Construction start Original/latest Original/latest Site Construction start Original/latest Original/latest
completion date cost estimate completion date cost estimate
Table 2). A new Chinese company, State Nuclear Power Technology periods – 12–15 years from start of construction to commercial opera-
Corporation (SNPTC), was set up in 2007 to participate in the con- tion – and their reliability has been mediocre, the three Konvoi plants
struction of imported AP1000s with a view to transferring the tech- have operated outstandingly reliably and their construction time was
nology to SNPTC so that the AP1000 could form the basis for China's about six years, longer than predicted for Gen III+ but better than
home market. It appears that hopes that costs would be reduced were many plants. In 2002, the joint venture was converted into a division,
not realised and SNPTC scaled up the AP1000 to form the CAP1400, Areva NP (66 per cent Areva, 34 per cent Siemens), of the newly formed
which SNPTC claims is their own intellectual property. Construction on unified French nuclear company Areva. Siemens withdrew from Areva
two pairs of reactors in the USA, Summer and Vogtle, started in 2013. NP in 2010 by which time Areva was 87 per cent owned by the French
The AP1000 received DAC in the UK in March 2017 and the developers state.
forecast that construction will start in 2020 on three AP1000s at the Construction of the first EPR, Olkiluoto (Finland) started in 2005,
Moorside site although the bankruptcy of Westinghouse means this is followed by Flamanville (France) in 2007 and two units in China
unlikely. KEPCO has been named as preferred bidder for the consortium (Taishan) in 2009/10. A consortium led by Electricité de France plans
owned by Toshiba for the Moorside site and if its bid is successful, it is to build two EPRs at the Hinkley Point site and it hopes to start con-
likely KEPCO would try to build its APR1400 technology. There are no struction in 2019. There are no other immediate order prospects.
other immediate order prospects. Thomas (2015) showed that all EPRs under construction are badly
Thomas (2015) shows that all the AP1000s are substantially delayed delayed and where costs are known, grossly over-budget (see Table 3).
and the US plants are far over budget. The pattern of problems appeared However, unlike the AP1000 the problems relate more to on-site
to be related to problems in the supply chain with the Reactor Coolant quality. The issues of poor quality concrete, welds, and steel, are typical
Pumps (RCP) a particular problem. The US customers began taking legal of the problems that previous design generations suffered. This reflects
action against Westinghouse soon after construction started. the fact that EPR is an updated, more complex version of existing de-
In October 2015, in response to a view that the problems with the signs.
AP1000 were in large part due to failures, particularly of quality control, The problems for the EPR were compounded by financial and
with its equipment suppliers, Westinghouse took over the nuclear busi- quality control problems with Areva NP that became apparent in 2015.
ness of its major supplier Chicago Bridge & Iron (CB&I) including Stone & In March 2015, Areva declared losses for the fifth consecutive year, of
Webster, an architect engineering company and the Shaw Group.20 €4.8bn and it became clear that Areva could not continue to trade
In October 2015, it renegotiated the Engineering Procurement and without powerful support from the French state. The proposed rescue of
Construction (EPC) contracts for its Vogtle and Summer projects on a Areva is based on splitting it into its component parts, the fuel cycle
fixed price basis to settle the legal action. A year later, it admitted that business and the reactor supply and servicing business. By April 2017,
the fixed price contracts underestimated costs by at least US$6.1bn. As the rescue of the fuel cycle business appeared on course, but the rescue
a result, the Westinghouse nuclear division was cast adrift from the of the reactor business was more problematic because of the huge po-
Toshiba main group, put into Chapter 11 bankruptcy protection and is tential liabilities it faced. In December 2018, EDF announced it had
for sale. In July 2017, the Summer project was abandoned. The Vogtle agreed to take-over of 75.5 per cent of the shares for €1.87bn in a re-
project is expected to continue. In January 2018, a Canadian private launched reactor vendor division reverting to its former name
equity group, Brookfield, announced it had agreed to buy Westinghouse Framatome.22 Mitsubishi Heavy Industries took 19.5 per cent of the
for US$4.6bn with completion expected in the third quarter of 2018.21 shares and a French engineering company, Assystem took the re-
If the deal is completed, it remains to be seen whether Brookfield will maining 5 per cent.
continue to market the AP1000. It appears the French government will take responsibility for his-
toric liabilities, particularly the cost over-runs at Olkiluoto and those
arising from defective components supplied or components with in-
3.2. EPR
adequate quality control documentation.23 In April 2015, the French
nuclear safety regulator, Autorité Sûreté Nucléaire (ASN), announced
The EPR had its origins in a joint venture between Framatome and
that the reactor bases and lids Areva had supplied from its Creusot forge
Siemens in 1992, which aimed to produce a design that could be li-
to the Flamanville and Taishan reactors did not meet specification with
censed across Europe, the European Pressurised Reactor (EPR). It was
based on the companies’ most recent designs, the Framatome N4 and
the Siemens Konvoi. While the four N4s had very long construction
22
https://www.edf.fr/en/the-edf-group/dedicated-sections/journalists/all-
press-releases/signing-of-definitive-binding-agreements-for-the-sale-of-areva-
20
Nuclear Intelligence Weekly ‘Westinghouse's Strategy in CB&I Stone & np-s-activities (Accessed January 19, 2018).
23
Webster Acquisition’ Oct 30, 2015. http://www.new.areva.com/EN/news-10873/edf-and-areva-sign-binding-
21
http://www.world-nuclear-news.org/C-Brookfield-to-buy-Westinghouse- agreements-for-the-sale-of-areva-np-s-activities.html (Accessed January 19,
for-4.6-billion-0401188.html (Accessed January 19, 2018). 2018).
219
S. Thomas Energy Policy 125 (2019) 216–226
Table 4
Record of the ABWR.
Source: https://www.iaea.org/PRIS/CountryStatistics/CountryDetails.aspx?current=JP (Accessed Mar 20, 2017).
Site Vendor Construction start/Commercial operation Lifetime load factor to end 2010 (%)
Note: Load factor is calculated as power produced as a percentage of the power that would have been produced had the plant operated uninterrupted at full design
rating.
Table 5 disaster in 2011. The two under construction in Taiwan have been
Record of the APR1400. mothballed since 2013 and are unlikely to be completed and it is not
Source: https://www.iaea.org/PRIS/CountryStatistics/CountryDetails.aspx? clear whether the two under construction in Japan will be completed.
current=JP (Accessed Mar 20, 2017).
The design was submitted to the US NRC by GE, Hitachi and Toshiba
Site Construction start Commercial operation and received Design Certification in 1997. However, the purchase of
Westinghouse by Toshiba in 2007 brought an end to the three-way
Shin Kori 3 (Korea) 10/08 12/16
collaboration. Hitachi and GE formed two joint ventures, Hitachi-GE
Shin Kori 4 (Korea) 08/09 –
Shin Kori 5 (Korea) 04/17 (80 per cent Hitachi) to operate outside the USA and GE-Hitachi (80 per
Shin Hanul 1 (Korea) 07/12 – cent GE) mainly for activity in the USA. GE-Hitachi had no serious sales
Shin Hanul 2 (Korea) 06/13 prospects by January 2018 while Hitachi-GE's only prospect was sales
Barakah 1 (UAE) 07/12
of two ABWRs to the UK for the Wylfa site with construction start
Barakah 2 (UAE) 04/13
Barakah 3 (UAE) 09/14
forecast for 2020.
Barakah 4 (UAE) 07/15 The ABWR is presented as a proven design, but all the construction
and operating experience is with a 30-year old version that does not
meet the criteria for Gen III+. Construction of the four completed
too much carbon in the steel.24 In October 2017, ASN approved op- ABWRs went smoothly but the operating performance has been poor
eration of the Flamanville plant only on the condition that the closure (see Table 4). How far this experience would be representative of the
head by replaced in 2024.25 By January 2018, the Chinese safety au- current version is difficult to determine.
thorities had not given an opinion on whether the Taishan vessels were
strong enough to allow operation.
Because of this, Areva is reviewing its records at all three of its 3.4. APR1400
factories going back to 1965. An initial review covering 9000 records at
Creusot found 400 irregularities, often falsifications, for equipment The Korean Electric Power Company's (KEPCO) Advance Power
such as reactor vessels and steam generators. This equipment has been Reactor 1400 (APR1400) is based on the System 80+ reactor licensed
installed not only in France but other countries including the UK, USA, from Combustion Engineering. KEPCO applied to the US NRC for
China, Japan and Switzerland. The French prosecutor is examining Design Certification in December 2014 even though it had no pro-
bringing criminal charges against Areva.26 In October 2016, the Pre- spective US customers.
sident of ASN, Pierre-Franck Chevet stated: ‘this "purge" of doc- One reactor of this design is complete and four more are under
umentation irregularities would continue. There is still one to two construction in Korea. In 2010, KEPCO won its first export order, for
years’ work. We will find other irregularities. It is obvious.’ four APR1400s to be supplied to the UAE. The winning bid was spec-
tacularly low, equating to about US$3600/kW, about 30 per cent lower
3.3. ABWR than the next cheapest bid (for EPRs). The Areva CEO, Anne
Lauvergeon was highly critical of the safety features in the APR1400
GE, Hitachi and Toshiba were long-term collaborators in the de- claiming it was like a car ‘without airbags and seatbelts’.27 KEPCO ac-
velopment of GE's BWR design for the Japanese market. The plan to knowledged the design did not contain expensive features that would
produce an Advanced Boiling Water Reactor (ABWR) design was an- be required in Europe such as a double containment to protect against
nounced in 1980 and the first orders placed in the late 1980s. Four aircraft impact and a core-catcher.28
reactors were completed in Japan with a further two under construction Construction of the first two units of this design was delayed due to
in Japan and two more in Taiwan using this early design. The four in the discovery of falsification of quality control documents for more than
Japan have produced minimal amounts of power since the Fukushima 2000 components (see Table 5).29 As with the ABWR, it is difficult to
evaluate the APR1400, which in the form it has been built does not
meet the criteria for Gen III+.
24
http://www.french-nuclear-safety.fr/Information/News-releases/ A further complication is whether KEPCO will get the support of the
Flamanville-EPR-reactor-vessel-manufacturing-anomalies (Accessed March 22,
2017).
25 27
http://www.french-nuclear-safety.fr/Information/News-releases/ Nucleonics Week ‘No core catcher, double containment for UAE reactors,
Flamanville-EPR-reactor-ASN-issues-its-opinion (Accessed January 19, 2018). South Koreans say’ April 22, 2010.
26 28
http://www.french-nuclear-safety.fr/Inspections/Supervision-of-the-epr- Nucleonics Week ‘South Korea preparing new designs for home, Europe,
reactor/Anomaly-affecting-the-Flamanville-EPR-reactor-vessel/Falsification-of- based on APR1400′ April 22, 2010.
29
materials-analysis-reports-ASN-is-collaborating-with-the-ongoing-judicial- http://www.world-nuclear-news.org/RS-Indictments_for_South_Korea_
inquiry (Accessed March 22, 2017). forgery_scandal-1010137.html (Accessed April 5, 2017).
220
S. Thomas Energy Policy 125 (2019) 216–226
Table 6
Russia nuclear orders post-1986.
Source: IAEA PRIS reactor data base: https://www.iaea.org/PRIS/home.aspx (Accessed September 8, 2016).
Country Site Technology No of units Construction start Commercial operation Construction time m’ths Load factor
Notes:
1. For reactors not yet complete but claimed to be within 2 years of completion, the construction time is estimated from the most recent estimate.
2. Includes only reactors with output greater than 150 MW and on which started construction after 1986.
3. Construction of Baltic 1 was suspended in 2013.
Korean government. This was an important element in winning the UAE placed for the home market orders and only four exports, two each to
order. The Korean President was elected in May 2017 with a promise to China and India, were won from 1986 to 2007 using designs, AES-91
phase-out nuclear power.30 It therefore seems questionable whether the and AES-92 respectively, with some (such as a core-catcher) of the Gen
Korean government will put its weight behind Korean reactor exports. III+ features (see Table 6). The plants for China took seven or more
There are no immediate prospects for orders for APR1400. years to complete, much longer than contemporary plants supplied by
Chinese vendors. The plants for India took 12–15 years to complete.
4. China and Russia In 2005, all parts of Russian nuclear industry were consolidated into a
new company, renamed Rosatom in 2007, with the backing of Vladimir
There is an increasing perception that if there is a market for re- Putin who installed a key ally, Sergey Kiriyenko, as CEO. In 2006, the
actors in the next decade or two, it will be supplied by Chinese and AES-2006 PWR design was announced which was claimed to meet all Gen
Russian vendors. Assessing the record of the Chinese and Russian nu- III+ requirements including a core-catcher, passive safety and aircraft
clear industries is difficult. There is little independent, authoritative protection. Four orders were placed for the home market with an ex-
information on reactor construction and operation, the regulatory pectation that home ordering would continue at a rate of about three
bodies are not transparent and there is no reliable cost information. reactors per year. However, by 2017, only three more PWRs had been
Neither country has experience of exporting reactors to countries, ordered for the home market, two using a pre-Chernobyl design while the
which, at the time of construction were open, had strong independent order for an AES-2006 was for the Kaliningrad enclave and construction
regulatory bodies and provided reliable costs.31 Information is there- was suspended in 2013 after only a year of work and is unlikely to re-
fore often fragmented and sometimes anecdotal. start. There were projections in 2017 of eight orders being placed by
The perceived advantages China and Russia are: 2025, but there must be doubts whether many of these will materialise.
The St Petersburg and Moscow design offices of Rosatom produced
• Designs claimed to meet the requirements of European and US their own significantly different versions of what is generally seen as a
regulators; single design. The Moscow office produced the AES-92 and their ver-
• Ability to provide finance; sion of the AES-2006 used a reactor design designated V-392M. The St
• Cheaper than other suppliers; Petersburg office produced the AES-91 and their version of the AES-
• Designs not tainted by escalating costs and construction delays; 2006 used the V-491 reactor. In 2010, Rosatom announced a new de-
• A viable home market to prove new designs; and sign, VVER-TOI produced by the Moscow office, which would super-
• Government backing of reactor sales. sede the AES-2006 with 20 per cent lower costs and a construction time
of only 40 months. However, by 2018, there were no orders for this.
In the next section we look at how justified these perceptions are. Export orders were won in large numbers from 2010 onwards and
by 2018, about 25 orders for AES-2006 – Belarus, India, Turkey,
4.1. Russia Bangladesh, Vietnam, Finland, Hungary and Egypt – had been placed
along with six for older designs – China, Iran and Jordan (see Table 7).
Russia has more than 60 years’ experience of designing and sup- Of these, construction had only started on the reactors in China, Belarus
plying power reactors. Up to 1986, it had completed 25 reactors for the and, in December 2017, Bangladesh. The orders for Vietnam and
Russian market with a further eight under construction and subse- Jordan now appear unlikely to go ahead.
quently completed. Its reactor exports were to former Soviet Republics The four AES-2006 reactors under construction in Russia seem likely
and to Comecon countries. Two reactors were exported to Finland in to take about 9 years to build. There is little information on what has
the 1970s but these incorporated Siemens instrumentation & control caused these delays (Thomas, 2015) but Russia's Audit Chamber put the
systems and Westinghouse containments and are therefore not re- blame on shortage of funds in a report from January 2015.32 However,
presentative. there have been reports of poor quality work and materials, and of
Following Chernobyl and from 1987 to 2007, no new orders were corruption. For example, in July 2011, steel structures for the con-
tainment building at the Leningrad site collapsed requiring 1200 t of
30
reinforcing steelwork to be dismantled to rebuild the destroyed wall of
Nuclear Intelligence Weekly ‘KHNP Slows Newbuild Design Work’ May 26,
2017.
31
Two Russian reactors were exported to Finland in the 1970s but these in-
32
corporated Westinghouse containments and Siemens instrumentation and Nuclear Intelligence Weekly ‘Auditor Report Illuminates Rosatom's
control systems and are therefore not representative. Financial Challenges’ January 23, 2015.
221
S. Thomas
Table 7
Russia's order book in 2016.
Source: Author's research. Note: Includes only reactors on which construction had not started by September 2016.
Country Site Unit Technology Original start to on-line Status Expected completion Agreed cost Russian finance
222
Nigeria Geregu/Kogi 2 2016–2025 Deal not complete 2025 US$8bn ?
S Africa Several 8 1200 MW −2022–29 No deal confirmed 2026 US$50bn Yes
S Arabia Not known 16 Not known −2030 No deal confirmed US$100bn No
a
Nucleonics Week ‘Akkuyu plant construction to begin in 2011, says Turkish energy ministry’ May 27, 2010.
b
Nuclear Intelligence Weekly ‘Moscow Meeting Breathes New Life into Akkuyu’ July 29, 2016, pp 4–5.
c
Nucleonics Week ‘Bangladesh, Russia initial contract for construction of Rooppur’ December 17, 2015.
d
Ibid.
e
Nucleonics Week ‘Russian industry to build Vietnam's first nuclear plant’ April 29, 2010.
f
Prime Tass ‘Rosatom to start designing Vietnamese nuclear plant in 2013′ November 12, 2012.
g
Power in Europe ‘Fennovoima aims for 2024′ September 16, 2013.
h
ITAR/TASS ‘Iran to fund construction of 2 new nuclear power units in that country’ November 11, 2014.
i
Nucleonics Week ‘Russia financing new units at Hungary's Paks’ January 16, 2014.
j
Nucleonics Week ‘Hungary to go ahead with Paks II plan despite EU concerns: government’ November 26, 2015.
k
BBC Monitoring Middle East ‘Jordan's first nuclear reactor to start operating by 2025 – official’ March 20, 2016.
Energy Policy 125 (2019) 216–226
S. Thomas Energy Policy 125 (2019) 216–226
China's widespread adoption of nuclear power came much later CNNC is the most established of the three because of its history and its
than that of Russia with a large construction programme launched in position appeared to be reinforced in 2017 when it announced its inten-
2008. It has three reactor vendors: tion to merge with China National Engineering Corp (CNEC), the effective
monopoly constructor of reactors in China. The position of SNPTC was
• China National Nuclear Corporation (CNNC) has its roots in the strengthened in 2013 when it merged with one of the five large Chinese
1960s in military nuclear applications; electric utilities, China Power Investment Corporation to form SPI.
• China General Nuclear (CGN) was founded in 1994 from the orga- From 2008 to 10 construction work started on 25 reactors in China,
nisation set up to participate in the construction of the first large six of which were the imported AP1000s and EPRs, two used an in-
reactors built in China imported from France; and digenous 650 MW PWR design, but the remaining 17 used the design,
M-310, designated CNP-1000, built in China in the 1990s and licensed
by Areva NP to CGN and CNNC. After the Fukushima disaster, there was
33
Bellona ‘Corruption: A new Russian Fukushima in the making?’ September
a marked slowdown in construction with existing construction delayed
27, 2011. http://bellona.org/news/russian-human-rights-issues/access-to- and from 2011 to the end of 2017, construction started on only 13
information/2011–09-corruption-a-new-russian-fukushima-in-the-making reactors with little sign that the previous pace of construction would
(Accessed February 27, 2015). return. These orders were split between five technologies: three CNP-
34
Bellona ‘Rosatom-owned company accused of selling shoddy equipment to 1000s, two AES-91s imported from Russia, four CGN ACPR-1000s, two
reactors at home and abroad, pocketing profits’ February 28, 2012 http:// CGN Hualong Ones and two CNNC Hualong Ones.
bellona.org/news/nuclear-issues/nuclear-russia/2012–02-rosatom-owned- CGN and CNNC began to produce their own advanced designs in
company-accused-of-selling-shoddy-equipment-to-reactors-at-home-and- 2011, ACPR-1000 and ACP-1000 respectively, building on the M310
abroad-pocketing-profits (Accessed February 27, 2015. rather than the EPR. Four of the ACPR-1000 design have started con-
35
Nucleonics Week ‘Rosatom says Belarus vessel undamaged in incident’
struction in China and two of the ACP-1000 are under construction in
August 11, 2016.
36 Pakistan.39 However, in 2013, the Chinese government required CGN
http://www.world-nuclear-news.org/NN-Russia-installs-RPV-at-Belarus-
plant-03041701.html (Accessed April 6, 2017). and CNNC to merge their designs to form Hualong One. Since then four
37
Channel News Asia ‘Putin removes head of VEB state development bank as
crisis bites’ February 18, 2016.
38 39
Nuclear Intelligence Weekly ‘Finland: Dearth of Qualified Personnel Stalls Pakistan has ordered 6 reactors from CNNC dating back to 1993, four using
Hanhikivi’ February 24, 2017, p 6. a Chinese 300 MW PWR design and two using the ACP-1000 design.
223
S. Thomas Energy Policy 125 (2019) 216–226
Table 9 before about 2024. This failure to win firm export orders is surprising
Construction time and operating performance of CNP-1000 reactors by year. given the success of Chinese industry in almost all other markets.
Source: IAEA PRIS database. https://www.iaea.org/PRIS/home.aspx (Accessed It seems clear that China's home market does give it the supply
January 22, 2018). chain to support many reactor exports. The China Development Bank
Year construction No of Construction time Lifetime load (CDB) and the Export & Import Bank of China are supporting state-
start units mths factor (%) backed companies, with CDB offering government- to-government low
interest loans to Argentina and Algeria for their nuclear programmes as
2005 1 57 84.1 (1)
2006 1 62 87.5 (1)
well as loans to CGN for the UK's Hinkley Point project. The Industrial
2007 1 70 79.2 (1) and Commercial Bank of China has agreed to offer loans of €10billion to
2008 6 68 88.3 (3) support CGN's nuclear project in Romania (Yu, 2015).
2009 5 75 –
2010 6 67 –
2012 2 54 – 4.3. Strengths and weaknesses of the Chinese and Russian nuclear industries
224
S. Thomas Energy Policy 125 (2019) 216–226
Table 10
Strengths & Weaknesses of the Russian and Chinese nuclear industries.
Source: Author's analysis.
Russia China
Finance Political support but doubtful capability Political support, strong apparent capability but untested
Design capability Long-established Little tested
Ability to satisfy experienced, independent Little tested Untested
regulator
Export order book Larger than it can handle No firm orders
Supply chain Weak, quality untested in open markets Strong, quality untested in open markets
Costs Little evidence but Finland & Hungary suggest No evidence
comparable to Areva/Toshiba
Home market Weak, consistently overestimated Potentially strong but issues of siting, over-capacity, technology choice
Construction record Mostly poor especially with AES-2006, construction Good with old design, experience with imported modern designs poor,
mishaps new designs untested, concern about quality
Industry structure Consolidated into one massive but unwieldy company Split between three bitter rival companies
5.1. Simplification and passive safety back loop from experience to improved designs is long. For example,
the generation of reactors that embody the lessons from Chernobyl is
Simplification would lead to lower costs. However, of the seven only beginning to enter service 30 years after the event.
designs that have made some progress towards ordering, six are basi- Experience with standardisation is not encouraging. Only two re-
cally previous designs with added safety features. It is implausible that actors using the SNUPPS design were built and their record of con-
adding systems and layers of defence can simplify the design and ex- struction and operation is far from outstanding. The French programme
perience with the EPR suggests the design is more complex and more of 58 reactors was split into seven ‘tranches’, covering three different
difficult to build than its predecessors. The more radical design, AP1000 output sizes with design changes between each tranche. Far from de-
relies for safety on passive features rather than active safety systems creasing over time, the real cost of these increased significantly and the
and this was the justification for the claim of lower costs. Experience final tranche, the N4 design, had a very poor construction record with
with the construction of the AP1000 is no better than that with the EPR the four reactors taking 11–16 years to complete (Grubler, 2010). There
and where it has competed in a tender with the EPR, it does not appear is little evidence that standardisation is an effective way to reduce costs
to have bid lower.43 and control construction times.
5.2. Modularisation
6. Where now for reactor technology?
Modularisation was one of the main factors in the claim that plants
All the ideas to save LWR technology appear to have been embodied
could be built quicker and with more predictability by moving con-
in Gen III+ and far from reducing costs and uncertainty in construc-
struction work from the site to a more controllable environment in a
tion, these appear to have been made worse. The cost gap between
factory. As with passive safety, the six designs evolved from existing
nuclear power and natural gas is now huge and, in many cases, re-
ones have minimal modularisation, so the only experience is with
newable costs are lower than nuclear, and, unlike nuclear costs, they
AP1000, which is poor. Quality control problems appear just to have
are continuing to fall in real terms. It also appears unlikely that reactors
been moved from the site to the factory with no net gain.
from Russia and China will solve the issues of cost and uncertainty in
construction. This implies that while the existing stock of LWRs might
5.3. Standardisation continue in service for several decades, they will not be replaced by
improved LWRs. So, if the nuclear industry is to have a future, it would
The nuclear industry has claimed standardisation would bring large appear to be either in more radical redesigns, so-called Gen IV reactors
economic benefits since the early 1970s when the Standardised Nuclear or Small Modular Reactors (SMRs). A full evaluation of these options is
Unit Power Plant System (SNUPPS) was proposed for five reactors in beyond the scope of this paper, but it is useful to outline the main ar-
the USA and the French nuclear power programme was launched. Those guments
advocating standardisation ignore two basic questions. What do they
mean by standardisation and why if it is such a sensible strategy has it
not occurred? 6.1. Gen IV
Standardisation can cover a range of models from the production of
essentially identical units – the Model T Ford example – to a more In 2000, a new international organisation, the Gen IV International
functional standardisation – for example, the QWERTY keyboard. While Forum (GIF) was set up to stimulate R&D in a new generation of nuclear
the QWERTY model has operational advantages it is hard to see how it designs that would be radically different to existing designs.44 By 2016,
would reduce costs. For the Model T model to work, the technology there were 14 partners including USA, UK, China and Russia. Six re-
needs to be so mature that standardisation does not block the in- actor technologies were selected for further development. Two, the
troduction of important technical progress or the incorporation of ex- Very High Temperature Reactor (VHTR) and the Sodium-cooled Fast
perience and the volume of sales needs to be large enough that the Reactor (SFR), were based on concepts that had already been built in
benefits of standardisation are reaped before there is a need to move on prototype and demonstration form, while the other four were new
to a more advanced design. Experience with nuclear power is still concepts. The two established concepts both have a 60-year history of
evolving rapidly, sales volumes are likely to remain low, and the feed- development and despite several independent attempts in about five
separate countries to commercialise these technologies, they have
proved problematic. The other four technologies will need significant
43
The AP1000 was chosen over the EPR in China but this appeared to be in
part due to Westinghouse's greater willingness to transfer technology to China.
44
Nucleonics Week ‘Westinghouse may win China bid as Areva balks at tech https://www.gen-4.org/gif/jcms/c_9335/charter (Accessed April 11,
transfer’ March 16, 2006, p 15. 2017).
225
S. Thomas Energy Policy 125 (2019) 216–226
technical advances for them to progress. The more expensive phase of supply such sensitive equipment. There is little evidence whether these
building and operating demonstration plants will only go ahead with designs would satisfy experienced and independent regulators.
large public subsidies and even if these are available there is no guar- Others take a more positive view on the future for nuclear power.
antee that commercially viable designs will be developed For example, the World Energy Council (2016), notes strong expansion
of nuclear capacity in the Far East and the potential of new designs,
6.2. SMRs Generation IV, SMRs and reactors designed for desalination. However,
since this report was published, governments committed to phase out
Small Modular Reactors (SMRs) have been talked about for more nuclear have been elected in Korea and Taiwan, the main priority for
than a decade. They can be divided into two categories: those based on Japan appears to be returning its existing reactors to service with new
scaled-down LWRs and those based more radical reactor concepts. It is orders a distant prospect and the flow of new construction starts in
hard to see how scaling down LWRs would reduce the cost per unit of China appears to have dried up. There are no commercial orders for
capacity over large LWRs, it seems more likely to raise the cost given Generation IV reactors (which mostly do not use LWR technology),
the loss of scale economies. As with Gen IV, these will require major SMRs or desalination plants.
development work and without large government subsidies, these are Many governments appear to have an inexhaustible capacity to
unlikely to be pursued by private companies. For a review of SMR de- discount past failed promises from the nuclear industry and are pre-
velopments see Ramana and Ahmad (2016) and Cooper (2014). pared to keep giving the nuclear industry another chance to prove itself.
This willingness has serious implications for other technologies which
7. Conclusions and policy implications will inevitably be pursued less vigorously if the assumption remains
that nuclear will provide the answers.
There is mounting evidence that Gen III+ reactors and the Nuclear
Renaissance they were meant to drive have failed. Far from reducing References
costs and uncertainty in construction, these appear to have been made
worse. The cost gap between nuclear power and natural gas is now huge Bupp, I.C., Derian, J.-C., 1978. Light Water: How the Nuclear Dream Dissolved. Basic
and, in many cases, renewable costs are lower than those of nuclear Books.
Cooper, M., 2014. The economic failure of nuclear power and the development of a low-
and, unlike nuclear costs, they are continuing to fall in real terms. It carbon future: Why small modular reactors are part of the problem not the solution.
appears unlikely that reactors from Russia and China will solve the is- Nucl. Inf. Resour. Cent. 〈https://www.nirs.org/wp-content/uploads/reactorwatch/
sues of cost and uncertainty in construction. newreactors/cooper-smrsaretheproblemnotthesolution.pdf〉 (accessed 11 April
2017).
The promise that measures such as rationalisation of safety systems, Grubler, A., 2010. The costs of the French nuclear scale-up: a case of negative learning by
use of passive safety and modularisation would dramatically reduce doing. Energy Policy 38.9, 5174–5188. https://doi.org/10.1016/j.enpol.2010.05.
costs and the risk of construction time overruns now appears to have 003.
Koomey, J., Hultman, N., Grubler, A., 2017. A reply to “Historical construction costs of
been an illusion. All the ideas to save LWR technology appear to have global nuclear power reactors”. Energy Policy 102 (3), 640–643.
been embodied in Gen III+. It is hard to see what other options remain Lovering, J.R., Yip, A., Nordhaus, T., 2016. Historical construction costs of global nuclear
to take LWR technology out of the spiral of increasing costs and com- power reactors. Energy Policy 91 (4), 371–382.
Lovering, J.R., Yip, A., Nordhaus, T., 2017. Apples and oranges: comparing nuclear
plexity that have made it commercially unattractive.
construction costs across nations, time periods, and technologies. Energy Policy 102
In the past, the stock of existing reactors has provided a buffer (3), 650–654.
giving a steady flow of servicing, maintenance and repair work to keep Ramana, M.V., Ahmad, A., 2016. Wishful thinking and real problems: small modular
reactor vendors going. However, the reactor servicing and maintenance reactors, planning constraints, and nuclear power in Jordan. Energy Policy 93,
236–245.
capability does not require a reactor vendor capacity. The two largest Thomas, S.D., 2015. Nuclear Construction Problems' PSIRU. University of Greenwich,
OECD reactor vendors, Westinghouse and Areva, are already in des- London. 〈http://www.psiru.org/reports/nuclear-construction-problems.html〉 (ac-
perate financial difficulties and if they survive, it may be only as cessed 14 June 2017).
Thomas, S., 2017a. China's nuclear export drive: trojan horse or marshall plan? Energy
maintenance, service and repair companies. Policy 101 (2017), 683–691.
The evidence available suggests that Chinese and Russian reactors Thomas, S.D., 2017b. Russia’s Nuclear Export Programme. PSIRU, University of
are not significantly cheaper or easier to build than their competitors. Greenwich, London. 〈https://www.gre.ac.uk/business/research/centres/public-
services/publications〉 (accessed 14 June 2017).
Russia seems unlikely to be able to provide finance on the scale re- World Energy Council, 2016. World Energy Resources: Uranium and Nuclear. World
quired to fulfil more than a small proportion of its order book while Energy Council. 〈https://www.worldenergy.org/wp-content/uploads/2017/03/
China has found it difficult to find customers. In both cases there are WEResources_Uranium_and_Nuclear_2016.pdf〉.
Yu, C.F., 2015. The IPO strategies of China's nuclear majors. Nucl. Intell. Wkly. IX
serious concerns about quality of components and work, safety culture
(40), 4–5.
and many countries would be reluctant to allow Russia or China to
226