Developing a range of levelized cost estimates for integral light water small modular reactors

Transcription

1 Developing a range of levelized cost estimates for integral light water small modular reactors Ahmed Abdulla and Inês L Azevedo 1 Department of Engineering and Public Policy, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA 1 Corresponding author: iazevedo@cmu.edu; tel.: +1 (412) Abstract The nuclear industry is developing a generation of small modular reactors (SMRs) with power outputs of up to 300 Megawatts-electric (MW e ). To control costs, vendors promise factory fabrication, shorter construction schedules, and increased use of modular construction. Their goal is to reduce the size of the plants initial capital outlay, to accommodate modest grids, and to allow for deployment in locations where large plants are infeasible. Interest has grown in incorporating SMRs into a portfolio of technologies that will reduce emissions from the power sector, and analysts are keen to assess their economics. Taking recent expert assessments of the cost of a 1,000MW e plant and two integral light water SMR designs, we calculate levelized cost of electricity values for four scenarios. For the large plant, median levelized cost estimates range from $56 to $120 per MWh. Median estimates of levelized cost range from $77 to $240 per MWh for a 45MW e SMR, and from $65 to $120 per MWh for a 225MW e unit. We compare these with other cost estimates from the literature. All values are in 2012 dollars. Controlling construction duration is important, and, given the price of electricity in some parts of the U.S., it is possible to construct an argument for deploying SMRs in some locations. Keywords: small modular reactors, levelized cost of electricity, nuclear power economics Classification codes: Q420; Q470; Q Introduction 1.1 Background information Nuclear power faced challenges well before the Fukushima nuclear accident. The events in Japan have sharpened arguments that have been in wide circulation throughout the technology s troubled history. These can be divided into four categories: safety of reactor operations; spent fuel stockpile management; possible diversion of fuel for weapons proliferation; and high capital cost. To address these concerns to differing extents, but especially high capital cost, the industry is developing a new generation of commercial small modular reactors (SMRs) that would have a power output of up to 300 Megawattselectric (MW e ), in contrast to the Gigawatt scale light water reactors (LWRs) that were favored by the industry in the nuclear renaissance of the 1960s [1] and continue to dominate it today [2]. Over the past few years, interest has grown in incorporating these small modular nuclear reactors into a portfolio of energy technologies that would reduce the carbon footprint of the power sector [3].

2 Therefore, utilities, energy analysts, and policy makers want to know how much they will cost. Three major challenges stand in the way of developing such cost estimates. First, the history of nuclear costestimation is poor. Section 2 below introduces some metrics used by the industry to assess economic viability (for a brief outline of nuclear power s record of cost-estimation, please consult appendix A). Second, SMRs encompass a large number of reactors that come in a wide range of sizes and employ a number of technologies. They are also slated for deployment in a variety of locations and for a number of applications. It is thus inappropriate to talk of likely SMR cost. One needs to discuss instead the likely cost of a given reactor design and its associated systems, as outlined in section 3. Third, given the fact that the detailed design of these reactors has yet to be completed, and that the dates of their commercial deployment are as of yet uncertain, existing approaches to estimating their likely cost are based on inappropriate or incomplete methodologies; these are highlighted in section 4. Taking recently published estimates of the likely cost of a 1,000MW e LWR and two integral light water SMR designs from a study that acknowledges the limitations noted above [4], section 5 complements existing estimates of SMR cost by calculating a range of levelized cost of electricity (LCOE) values for four proposed plant configurations. Section 6 presents our conclusions Recent large nuclear power plant cost estimates Two metrics of great interest when it comes to nuclear power are a plant s overnight capital cost and its levelized cost of electricity. The first of these is the cost of constructing a nuclear plant minus the cost of financing it over the construction period. The levelized cost of electricity is the price the nuclear plant must charge per unit of electricity sold for it to break even over its lifetime Overnight capital cost The emphasis on capital costs in the case of nuclear power stems from the facts that: construction costs make up a large fraction of the total cost of nuclear power [5]; their operating costs are low given the energy density of their fuel; and the dearth of recent construction experience renders the estimates more uncertain and the projects financially riskier. Vendors were optimistic at the turn of the century when generating estimates of the capital cost of new nuclear power plants. Initial estimates that the overnight cost of new plants would range between $1,000 and $2,000 per kw of capacity are obsolete [6, 7]. Table A1 in appendix A lists some of the capital cost estimates that have been published in the past Levelized Cost of Electricity (LCOE): Customers, policymakers, and energy analysts are interested in more than just nuclear power s capital costs. The LCOE is a more comprehensive measure that takes into account capital costs, both fixed and variable operating costs, decommissioning costs, the plant s construction duration, its lifetime, capacity factor, heat rate, and the cost of financing the project. Equation 1 below depicts one approach to computing the LCOE: LCOE =!!!!!"!!!! (!!!)!!!!!!! (!!!)! (1) where n I! OM! d E! is plant lifetime is the investment cost in year y (initial and incremental capital) is the operating and maintenance cost in year y (fixed and variable) is the discount rate is the electricity generated in year y 2

3 Another approach, described in detail for the case of concentrated solar power by Wagner and Rubin [8], is to compute the levelized annual cost (LAC) of a technology and divide that annual cost by the average annual electricity generated. These two methods using equation 1 above or the approach used by Wagner and Rubin give broadly similar results. For detailed commentary on methods of assessing the cost of electricity, please consult Kammen and Paca, 2004 [9]. One influential [5] study of nuclear power economics in the twenty-first century is MIT s Future of Nuclear Power [10]. In this study, MIT constructed a cost schedule for a hypothetical Gigawatt-scale nuclear plant and, applying the procedure presented in Equation 1, concluded that the LCOE from this plant would range from $42 to $67/MWh of electricity generated. In 2009, a team from MIT updated the figures to take into account changes in the commodities market, as well as revised estimates of the cost of nuclear power plants, estimating a LCOE of $84/MWh [11]. Although it is possible to criticize the structure of the cost schedule employed in these studies as too academic, the aforementioned lack of experience with nuclear construction makes studies by academic outfits important in the case of this technology. Table 1 lists some of the nuclear power levelized cost estimates made over the past decade. Generally, these studies depend on a number of assumptions about plant operating characteristics, investment parameters, and even government treatment of such investments for tax purposes. Given these differences in assumptions, comparisons of LCOE values must be made with care. Some of the pertinent assumptions made during the calculation of the LCOEs in table 1 are: first, the plants are assumed to have a capacity factor of either 85% (studies 1, 2, and 3) or 90% (studies 4, 5, and 6); second, estimates exclude subsidies that would make nuclear power more economically attractive to utilities; third, all estimates but the fourth include the cost of decommissioning the plant. Table 1. Recent estimates of the LCOE from large nuclear power plants. No. Year of estimate Source Year of $ LCOE ($/MWh) MIT [10] U Chicago [12] MIT [11] Lazard [13] EIA [14] U Utah [5] One column in tables 1 and A1 is reserved for the year in which the cost figure is reported. Updating the cost to current-year dollars is not a simple matter of accounting for inflation. Some of the assumptions (such as the capital cost and the fuel cost) depend on other indices. Vendors retain a suite of such indicators to account for changes in the cost of procurement, construction, and commodities, as well as indices of regional cost variation, when developing estimates of the cost of proposed new builds SMRs come in a wide range of sizes and technologies According to the International Atomic Energy Agency (IAEA), small reactors are reactors that produce less than 300MW e. Nine IAEA member states are engaged in developing small reactors: Argentina, Brazil, China, France, Japan, Russia, South Korea, and the United States. The Agency identifies twenty-five small reactor designs in development worldwide [15]. A brief description of these is provided in table B1 in appendix B, along with a description of two SMRs we are aware of that are not in the IAEA publication referenced. 3

4 Levelized cost estimates of light water small modular reactors The broadest division that can be made is between light water SMRs and non-light water SMRs. The world s existing nuclear power plant fleet mostly utilizes LWRs: 356 of the world s 436 nuclear reactors (80%) are of the light water type [2]. Because of the dominance of this technology, vendors, operators, and regulators have great experience with LWRs and, when they became interested in developing smaller units, established vendors opted to base their designs on this technology in an effort to leverage their experience while alleviating both customer and regulator concerns. These designs make more likely candidates for near- to mid-term deployment. More than half of the SMR designs listed in table B1 are of the light water variety. These can be further subdivided into conventional light water SMRs and integral light water SMRs. The former are scaled down version of the Gigawatt-scale reactors currently in operation, while the latter integrate the components of the nuclear steam supply system (NSSS) most notably the core, the steam generator, and the pressurizer into a single pressure vessel, eliminating much of the nuclear-grade piping. Figure 1 below compares the steam supply systems of conventional and integral light water reactors. Figure 1. Compare the nuclear steam supply system of a 1,150MWe, Gen III+ Westinghouse AP1000 (left) to that of a 225MWe Westinghouse SMR (right). The NSSS is integrated into one module in an integral light water SMR [16 17]. Vendors claim that these integral light water SMRs would be manufactured on a factory assembly line with high levels of quality control, after which they would be shipped to the site by road, rail, or barge. Other advantages they promise are shorter construction schedules and the increased use of modular construction techniques, both of which might control deployment costs. Because they come in smaller sizes, utilities may be able to purchase nuclear capacity in smaller increments to match demand growth, and the absolute cost of the investment would be lower, opening up the market to those utilities and organizations that cannot afford to bet the company to acquire a nuclear plant [18]. Finally, because of their smaller size, they permit novel approaches to siting, such as the co-siting of many modules [19], or underground [19 21] or underwater [22] deployment, that are infeasible for large reactors. Non-light water designs encompass a variety of technologies, all of which fall under the Generation IV label [23]. In the United States at least, these designs are destined for deployment only in the long-term because, among other reasons, they require more substantial changes to the framework governing deployment than light water SMRs do. That said, the advanced designs not only attempt to control costs in the ways light water SMRs do, but also promise to deal with nuclear power s other disadvantages by being safer, more resistant to the proliferation of nuclear materials, and more innovative in managing waste. Some designs even promise to deliver sealed modules that operate for long lifetimes and are ultimately disposed of, or sent to the factory to be recharged, once their fuel is exhausted [24]. 4

5 Even the basic summary above shows how difficult answering the question of likely SMR costs is. There is the issue of reactor type, reactor size, and the nature of the deployment (a one-module versus twelve-module plant, for example). The scope of any cost estimate needs to be carefully defined Existing approaches to estimating the cost of SMRs The few economic analyses of SMR cost in the public literature are problematic. Some of these studies have employed a top-down approach that estimated SMR cost by scaling down from the cost per kw e of large reactors (e.g. [25]). As the previous section emphasizes though, SMR designs are different from their larger cousins, with the sole arguable exception of small conventional LWRs. This places them on a different cost curve. Few studies have employed bottom-up engineering-economic assessments of SMR capital cost: these decompose an SMR into its major constituent components (many of which have yet to be fabricated) and build up a total capital cost estimate using a combination of authors judgments and consultation with component vendors (e.g. [26]). SMR vendors are conducting more robust bottom-up cost analyses, but their data are proprietary. Given that the size of the operating staff and maintenance plans, among other things required to deploy SMR plants of various sizes and configurations, have yet to be determined by vendors, let alone approved by any regulator, the few existing estimates of operating and maintenance (O&M) costs are almost entirely based on conjecture. In table 2 below, we list some existing estimates of SMR overnight cost from the literature. Table 2. Some existing estimates of SMR overnight cost, adjusted to 2012 dollars. No. Year of Overnight cost Source estimate ($/kw e ) IAEA Generic SMR [25] 4, Energy Policy Institute: Typical SMR [27] 5, Electric Power Research Institute: Generic estimate [28] 5,000 5, Nuclear Energy Agency: 4 PWR-335 [29] 4,900 5, Nuclear Energy Agency: 5 PWR-125 [29] 6,800 8, American Security Project: 100MW plant [30] 2, Energy Policy Institute: Typical SMR [31] 6, Anadon et al., expert elicitation, SMR cost in 2030 [32] 1,000 16,000 In a previous paper [4], we ran an expert elicitation in order to improve on the existing estimates of SMR capital cost. We argued that, when done properly, expert elicitations can complement [the] approaches [mentioned earlier]. We sat down with sixteen experts working directly or indirectly on SMR development, including fifteen experts who worked for nuclear technology vendors (as employees or contractors) and asked them for their assessment of the overnight cost of reactor deployment scenarios that involved a Gigawatt-scale current generation LWR and two integral light water SMRs. We developed technical descriptions of the two SMRs, a 45MW e unit and a 225MW e unit, based on publicly available data. Table 3 below outlines four of the scenarios we developed for the experts to analyze. Table 3. Four hypothetical nuclear reactor deployment scenarios posited to our experts. Number and type of reactors on the same site Individual reactor capacity (MW e ) Total plant capacity (MW e ) Scenario 1 1 typical Gen III+ PWR 1,000 1,000 Scenario 2 1 light water SMR Scenario 3 5 light water SMRs

6 Scenario 4 1 light water SMR We defined overnight cost as the cost in 2012 dollars of engineering, procurement, and construction (excluding owner s cost) and assumed that the plants in question were Nth-of-a-kind (NOAK). Other assumptions were made to craft a consistent U.S.-centric scenario; for more details, please consult [4]. For individual experts estimates of the overnight cost of each scenario, please consult table C1 in appendix C. As well as asking questions about overnight capital cost, we asked experts to estimate the construction duration of each of the single-unit plants (scenarios 1, 2, and 4 in Table 4 above). Construction duration was defined as the length of time from the pouring of first safety concrete to plant commissioning. There was consensus that the large reactor plant scenario 1 would require a construction duration of five years at NOAK. Integral light water SMRs were generally thought to require three years at NOAK. Table C2 in appendix C summarizes expert responses to this question. 2. Materials and methods 2.1. Building a construction schedule for the four deployment scenarios We constructed a cost schedule based on the Update on the Cost of Nuclear Power and, after reproducing the results of the MIT study [11], we use the data in tables C1 and C2 to generate estimates of the cost of each of the four scenarios. Table 4 lists some of the parameters used in constructing the cost schedules. Note how we retain some of the values used in the MIT study for the purpose of comparing our results to existing estimates. This is especially true where no better information on which to update the existing data exists. For our baseline scenario, we adopt a weighted average cost of capital (WACC) of 10%, a heat rate of 10,400 BTU/kWh, and a 37% tax rate the same assumptions made in the MIT report [11]. We change MIT s capacity factor from 85% to 90% and the lifetime of the plant from 40 to 60 years to reflect the fact that Gen III+ builds are designed to operate at these higher capacity factors and for this extended time period. The construction duration for the conventional reactor is modeled as a five-year schedule, with 10% of the construction performed in each of the first and fifth years, 25% of the construction completed in each of the second and fourth years, and 30% of the construction completed in year 3. This implies an S- shaped construction profile. The single-unit SMR plants take three years to build; the capital spent is spread out evenly among the years. The multi-module SMR plant (scenario 3) takes five years to build; the capital is spread out evenly among the years. Table 4. Parameters used to calculate the LCOE for the scenarios under investigation. See section 5.2 and the supplementary materials for an analysis of the sensitivity of LCOE to some of these parameters. Variable (Units) Scenario 1 Scenario 2 Scenario 3 Scenario 4 Capacity (MW e ) Capacity factor (%) Construction duration (years) Heat rate (BTU/kWh) Overnight cost ($/kwe) 1, ,400 10,400 10,400 10,400 Data derived from expert elicitations (table C1) 6

7 Incremental cap. Cost ($/kwe) Fixed O&M cost ($/kwe/year) Var. O&M cost (mills/kwh) Fuel costs ($/mmbtu) Waste fee ($/kwh) Decommissioning costs ($) O&M real escalation (%) Fuel real escalation (%) Tax rate (%) WACC (%) Lifetime (years) 1% of overnight cost Calculated from Code of Federal Regulations The overnight capital costs were derived from each individual expert s distribution, and the incremental capital cost was calculated from the overnight cost. The resulting incremental capital cost figures, especially for the single-unit SMR plants, may be too low. However, SMR vendors promise that these designs will require less maintenance. In any event, absent more information about how reliable these will be, there exists no basis on which we can update MIT s estimates. Similarly, because we never asked for O&M cost estimates in our elicitation procedure, the values used in our model were updated versions of those used by MIT. The scaling was done using the consumer price index. Rounding out our discussion of the remaining parameters in table 4, fuel costs were taken from the Nuclear Energy Institute s 2011 estimate of delivered nuclear fuel cost [33], the waste fee is set by statute, and the decommissioning costs were calculated using the Code of Federal Regulation s (CFR) current decommissioning funding allowance (DFA) requirements. 2.2 Computing LCOE We used an application of Equation 1 shown in section 2.2 to compute an average cost of electricity, which is the ratio between the discounted after-tax cash flows (the numerator in equation 2) and the discounted energy output: LCOE =!!!"#$%!!!!! C t construction Rtax C t depreciation + 1 Rtax C t o&m!!!"#$%!!!!!!"#$%!!"#$"#!!!!"#!!!!!"#!!!!"##!!!!!"##! (2) 7

8 Here, WACC is the weighted average cost of capital; R!"# is the rate of inflation; R!"# is the expected rate of taxation; Output (in kwh) is the annual electricity output, which is computed by multiplying the plant capacity S (in kw e ) by the capacity factor and by 8,760; and C!!"#$%&'!%("# is the construction cost in year t (in dollars per kw e, and is a function of construction duration and schedule). Since different types of plants have different construction profiles, we included these explicitly in the LCOE estimate, as table 4 above shows. While capital costs are incurred in the initial year, electricity generation only starts once the construction ends: t const is the difference between the year in which construction begins and the first year the plant generates power. Thus, n+tconst is the last year of plant operation. C!!"#$"%&'(&)* is the depreciation amount in year t (in dollars). Our depreciation schedule follows the Modified Accelerated Cost Recovery System, as per IRS regulations for large power plant projects. C!!&! is the cost of operating and maintaining the plant in year t (in dollars). This is composed of: C!"" C!"#$%% C!"#$&! C!"#$&! C!"#$ C!"#$% Incremental capital cost (in dollars per kw e per year) Decommissioning cost (in dollars) when t = n+tconst Fixed O&M cost (in dollars per kw e per year) Variable O&M cost (in dollars per kwh) Fuel cost (in dollars per mmbtu) Waste fee (in dollars per kwh) All these costs are subject to inflation and real cost escalations, except the waste fee, which is fixed by statute. Equation 3 below presents the components of the O&M cost variable: C!!&! = S C!"" + C!!"#$%% 1 + R!"#! + S C!"#$&! + Output C!"#$&! 1 + ESC!&! 1 + R!"#! + Output C!"#$% + Output HR C!"#$ 1 + ESC!"#$ 1 + R!"#! Three of the variables in Equation 3 were not defined above. These are ESC!&!, which is the percentage of O&M cost real escalation; HR, which is the plant s heat rate (in BTU/kWh); and ESC!"#$, which is the percentage of fuel cost real escalation. 3. Results and discussion Figure 2 below shows the levelized cost of electricity for the four scenarios, using the estimates of individual experts. All values are in 2012 dollars. (3) 8

9 Figure 2. Using individual experts overnight cost estimates, we calculate the LCOE in 2012 $/MWh of (1) a 1,000MW e Gen III+ reactor plant, (2) a 45MW e integral light water SMR plant, (3) a five-module 45MW e light water SMR plant, and (4) a 225 MW e light water SMR plant. WACC = 10%. Uncertainty ranges are provided for each sub-component of total system LCOE. There is no consensus among our experts regarding the overnight cost of either the large reactor or the three SMR scenarios. Naturally, given nuclear power s intensive capital requirements, the estimates of levelized cost therefore span a wide range. For the large plant, median levelized cost estimates range from $56 to $120 per MWh. Five of the sixteen estimates (K, M, N, O, and P) suggest a median LCOE greater than $100 per MWh for the large reactor plant; only two (A and B) suggest an LCOE less than $80 per MWh. The capital intensiveness of nuclear power is clear from figure 2: levelized capital cost accounts for anywhere from around 60% to 80% of total system levelized cost in scenario 1. Eight of the sixteen experts have median overnight cost estimates within 10% of the median overnight cost estimate of the aggregated expert distributions for this scenario. The wide range of SMR overnight cost estimates elicited from the experts leads to a wide range of levelized cost estimates for these scenarios too. Median estimates of system levelized cost range from $77 to $240 per MWh for scenario 2, with levelized capital cost again accounting for around 60% to 80% of system levelized cost. Twelve of the sixteen median estimates suggest a LCOE greater than $100 per MWh. Estimates of the median overnight cost vary considerably, with only one expert s median overnight cost estimate falling within 10% of the median overnight cost estimate of the aggregated expert distributions for this scenario. When co-locating five 45MW e SMRs on one site (scenario 3), median estimates of total levelized cost range from $81 to $230 per MWh. Levelized capital cost accounts for around 65% to 80% of total system levelized cost in this scenario, and five of the sixteen experts median overnight cost estimates fall within 10% of the median overnight cost estimate of the aggregated expert distributions for this scenario. 9

10 Although the overnight cost estimates of this scenario are generally lower than those of scenario 2, we assumed that it takes five years for the staggered construction of these five co-located units to be completed, compared with three years for scenario 2. Under current regulations, it is improbable that modules could be commissioned while adjacent modules remain under construction. Innovative construction-operation interfaces, if approved by the regulator, might change the economics of such deployments. Despite the longer construction duration, which leads to a delay in the initiation of the revenue stream, these units still have a LCOE lower than that of scenario 2, thanks to the economies of scale associated with the co-location of multiple modules on the same site, leading to lower overnight cost estimates in our experts judgment. The 225MW e SMR in scenario 4 generates estimates of total system levelized cost that are lower than those of the other SMR scenarios, and only slightly higher than for the large reactor. Despite its shorter construction duration (three vs. five years), its higher overnight cost still puts it at a disadvantage relative to the large reactor. Median estimates for scenario 4 s total system levelized cost range from $65 to $120 per MWh. Seven of the sixteen experts have median overnight cost estimates within 10% of the overnight cost estimate of the aggregated expert distributions for this scenario. To generate one estimate for each reactor type, we aggregate the assessments of the sixteen experts, assigning equal weights to each. Table 5 below summarizes our results: Table 5. Range of LCOE estimates for the four nuclear power plant deployment scenarios, aggregating the judgments of sixteen experts, each of whom is assigned equal weight. Scenario Mean WACC; Levelized cost of energy ($/MWh) overnight cost lifetime ($/kwe) (years) 5 th perc. 50 th perc. 95 th perc. 1: 1 1,000MW e 4,900 10%; : 1 45MW e 8,500 10%; : 5 45 = 225MW e 6,900 10%; : 1 225MW e 5,300 10%; The median LCOE estimate for the Gigawatt scale reactor is $91 per MWh. The small SMRs have a higher levelized cost. Locating one small (45MW e ) SMR on a site will yield an LCOE of $139 per MWh (median estimate), with the 90 th confidence interval ranging from $123 to $158 per MWh. Locating five small units on a site would be less expensive, owing to the lower overnight cost distribution. This despite the longer construction duration (five instead of three years) and the assumption that the plant can only be commissioned once all five units are in place. Again, scenario 4, a plant consisting of a single 225MW e SMR, yields only a very slightly higher LCOE estimate than the large reactor. Although its overnight cost is greater than the large plant, we assume that it is brought online two years faster than the large plant, generating a revenue stream sooner. The shorter construction schedule fails to neutralize the premium associated with the SMR s higher overnight cost. We conducted a sensitivity analysis on these results. Please consult appendices D and F for figures and tables demonstrating the effects of varying the WACC and the lifetime of the plants on total system levelized cost, and for a figure comparing these scenarios to alternative methods of electricity generation (figure F1). In figure 3, we compare the distribution of total system levelized costs generated in this paper with a distribution of U.S. electricity prices in 2011 [34]. All prices are in 2012 dollars. The distributions for the four plants represent the aggregate of expert assessments of overnight cost, along with the uncertainties in other LCOE parameters reported in table 4. The distribution of U.S. electricity prices takes statewide average prices across all sectors, weighted by volume of retail sales in each state. Although the LCOE distributions span a broad range, average electricity prices in some U.S. states are above the total system levelized costs for SMRs. Average electricity prices in California, New York, and Alaska, rounded to two significant figures, are $130, $150, and $160 per MWh, respectively. Hawaii s 10

11 electricity is the most expensive of any state, with an average price of almost $290 per MWh. SMRs, even at the high capital costs envisioned by some of our experts, might yet find viable applications. Figure 3. Comparing the distribution of total system levelized costs generated in this paper with U.S. electricity price in 2011 across all sectors. Even at the upper end of our estimate of SMR levelized costs, these reactors may find some economically viable applications in the U.S. All figures are in 2012 dollars. As figure 3 shows, scenarios involving the smaller SMR already cater to the tail end of the distribution of electricity prices in the U.S. Five-sixth of total electricity sales in the U.S. cost consumers less than $130/MWh. For the median LCOE of the Gigawatt-scale plant to hit that target price, the median overnight cost of the aggregated expert distributions would have to reach $7,600 per kw e as opposed to the current estimate of $4,900, an increase of more than half. The 225MW e SMR has only a slightly smaller margin for the LCOE to reach $130/MWh. However, the smaller SMR has no such margin: if sited as a stand-alone unit, its median LCOE is already higher than $130/MWh; if five units are co-sited, the margin is negligible: around 5% ($7,200 per kw e versus $6,900). We plug existing estimates of SMR LCOE those shown in table 2 above into our cost schedule to determine where our results stand relative to the existing literature. Table 6 summarizes the results of this exercise: Table 6. Estimates of SMR LCOE generated using overnight cost estimates from the literature. We assume that the one-unit plants take 3 years to deploy, similar to scenario 4, while the multi-module deployments take 5 years, similar to scenario 3. See table 4 for other assumptions made in these scenarios. No. Year of LCOE ($/MWh) Source estimate [5 th, 50 th, 95 th ] IAEA Generic SMR [25] 77, 81, Energy Policy Institute: Typical SMR [27] 90, 94, Electric Power Research Institute: Generic estimate [28] 87, 94, Nuclear Energy Agency: 4 PWR-335 [29] 92, 99, Nuclear Energy Agency: 5 PWR-125 [29] 120, 135, American Security Project: 100MW plant [30] 53, 57, 62 11

12 Energy Policy Institute: Typical SMR [31] 102, 107, Conclusions In this paper, we have sought to expand our understanding of where the economic argument for SMRs stands. We believe that controlling construction duration, while perhaps important from a financial risk management perspective, is not as important a factor as controlling overnight capital cost, given how sensitive the levelized cost is to changes in this parameter. Given the high price of electricity in certain locations, it is possible to construct an economically viable argument for deploying SMRs for some applications, though their deployment may come at a premium compared with other technologies. As appendix G in the supplementary materials argues, however, there are costs, incurred by both SMRs and alternative energy technologies, that the LCOE fails to account for, including potential environmental damage, security of fuel supply, and various socio-political, geo-political, and institutional constraints that may tip the decision in favor of one technology or another. The strategic business case for SMRs rests on factory fabrication, shorter construction schedules, and, for economies of volume to materialize, the existence of an international export market or a large domestic order book. Whether any of these, let alone all three, will materialize remains to be seen. Acknowledgements Ahmed Abdulla was supported by the Crown Prince s International Scholarship Program, Bahrain, and by the Steinbrenner Institute at Carnegie Mellon. Inês Azevedo was supported by the center for Climate and Energy Decision Making (SES ) through a cooperative agreement between the National Science Foundation and Carnegie Mellon University, and by a grant from the John D. and Catherine T. MacArthur Foundation ( INP). We would also like to thank the Carnegie Mellon Electricity Industry Center. References [1] Bupp IC and Derian JC 1978 Light Water: How the Nuclear Dream Dissolved (New York: Basic) [2] IAEA 2013 Power Reactor Information System ( [3] Chu, S 2012 Charge to the Secretary of Energy Advisory Board Small Modular Reactors (SMRs) Subcommittee: Memorandum from the Secretary of Energy (Washington, DC) [4] Abdulla A, Azevedo IL and Morgan MG 2013 Expert Assessments of the Cost of Light Water Small Modular Reactors Proc. Natl Acad. Sci. 110(24) [5] Hogue MT 2012 A Review of the Costs of Nuclear Power Generation (Salt Lake City, UT: Bureau of Economic and Business Research, David Eccles School of Business, University of Utah) [6] Cooper M 2009 The economics of nuclear reactors: renaissance or relapse? (South Royalton, VT: Institute for Energy and the Environment, Vermont Law School) [7] Kidd S 2009 Escalating costs of new build: what does it mean? (Nuclear Engineering International Magazine, available at: [8] Wagner SJ and Rubin ES 2012 Economic implications of thermal energy storage for concentrate solar thermal power Renew Energy, available at: [9] Kammen DM and Pacca S 2004 Assessing the Costs of Electricity Annu. Rev. Environ. Resour [10] Ansolabehere S, Deutch J, Driscoll M, Gray PE, Holdren JP, Joskow PL, Lester RK, Moniz EJ, and Todreas NE 2003 The Future of Nuclear Power (Cambridge, MA: Massachusetts Institute of Technology) [11] Du Y and Parsons JE 2009 Update on the Cost of Nuclear Power (Cambridge, MA) 12

13 [12] Tolley GS et al The Economic Future of Nuclear Power (Chicago, IL: The University of Chicago, available at: [13] Lazard 2011 Levelized Cost of Energy Analysis Version 5.0 (available at: [14] EIA 2012 Levelized Cost of New Generation Resources in the Annual Energy Outlook 2012 (available at: [15] IAEA 2012 Status of Small and Medium Sized Reactor Designs (available at: [16] Matzie RA 2003 The AP1000 reactor nuclear renaissance option (New Orleans, LA: presentation at the Tulane Engineering Forum) slide 9 [17] Nuclear Street News 2011 Westinghouse Unveils New Small Modular Reactor Design (available at: 18/westinghouse-unveils-new-small-modular-reactor-design aspx) [18] Rosner R and Goldberg S 2011 Small Modular Reactors Key to Future Nuclear Power Generation in the U.S. (Chicago, IL: The University of Chicago, available at: opy.pdf) [19] Reyes JN 2011 Overview of NuScale Technology (Vienna, Austria: presentation to the International Atomic Energy Agency) [20] Anness M 2011 Westinghouse SMR Product Development (Vienna, Austria: presentation to the International Atomic Energy Agency) [21] Kim TJ 2010 Generation mpower (Washington, DC: presentation to the American Nuclear Society) [22] DCNS 2012 Flexblue (available at: [23] Bouchard J and Bennett R 2008 Generation IV Advanced Nuclear Energy Systems Nuclear Plant Journal 26(5) [24] Gen 4 Energy 2013 Safety & Security (available at: [25] Kuznetsov V and Barkatullah N 2009 Approaches to Assess Competitiveness of Small and Medium Sized Reactors, presented at the International Conference on Opportunities and Challenges for Water Cooled Reactors in the 21 st Century, IAEA Paper 1S01 (Vienna, Austria: International Atomic Energy Agency) [26] Zhang Z and Sun Y 2007 Economic potential of modular reactor nuclear power plants based on the Chinese HTR-PM project Nucl. Eng. Des [27] Solan D et al Economic and Employment Impacts of Small Modular Nuclear Reactors (Boise, ID: Energy Policy Institute, Boise State University) [28] Welling C 2010 SMR Financing and Economics The Nuclear Option: Is Small Scale Nuclear Energy an Option for Alaska? (Fairbanks, AK: presentation at the University of Alaska Fairbanks) [29] Nuclear Energy Agency 2011 Current Status, Technical Feasibility, and Economics of Small Nuclear Reactors (Paris, France: Nuclear Energy Agency, available at: [30] Cunningham N 2012 Small Modular Reactors: A Possible Path Forward for Nuclear Power (Washington, DC: American Security Project, available at: %20Small%20Modular%20Reactors.pdf) [31] Black G 2012 Estimating the Economic Impacts of Small Modular Reactors (Arlington, VA: presentation to the Platts 3 rd Annual Small Modular Reactors Conference) [32] Anadon LD, Bosetti V, Bunn M, Catenacci M and Lee A 2012 Expert Judgments about RD&D and the Future of Nuclear Energy Environ. Sci. Technol. 46 (21)

14 [33] NEI 2011 Costs: Fuel, Operation and Waste Disposal (Washington, DC: Nuclear Energy Institute, available at: [34] EIA 2012 State Electricity Profiles 2010 (available at: 14