Download PDF Fall Issue of The Bridge on Nuclear Energy Revisited September 15, 2020 Volume 50 Issue 3 The desire to reduce the carbon intensity of human activities and strengthen the resilience of infrastructure key to economic prosperity and geopolitical stability shines a new spotlight on the value and challenges of nuclear energy. The Case for Nuclear as a Low-Carbon, Firm, Widely Available Energy Source Friday, September 18, 2020 Author: Karen Dawson, Michael Corradini, John Parsons, and Jacopo Buongiorno Deep decarbonization of economies will require thoroughgoing changes to all parts of the energy system, including replacing a large share of fossil fuel consumption with low-carbon sources. What will be nuclear’s place in this transformation? Options for Decarbonized Energy Nuclear power is the largest source of low-carbon energy in the United States and Europe, and the second largest source worldwide after hydropower. In the past, nuclear was primarily chosen as a baseload technology, evaluated in comparison against other baseload options such as coal- or natural gas–fired technologies. But will these be its competitors in a deeply decarbonized system? Wind and Solar The fastest-growing sources of low-carbon generation worldwide are wind and solar photovoltaic (PV) technologies. Over the years 2018–50, the US Energy Information Administration’s reference case scenario shows these technologies growing at average annual rates of 4.9 percent and 7.2 percent worldwide, respectively, while the growth rate for nuclear is approximately 1 percent (EIA 2019). The expanding grid penetration of renewables is changing the competitive landscape. But the variability of wind and solar energy resources injects a new dimension to the problem of choosing a portfolio of investments that reliably matches supply with demand at low cost. Even the modest penetration of renewables observed so far has forced changes to the operation of national grids to adapt to the high variability of wind and solar resources. Some of this adaptation has been just a matter of time and technical innovation. For example, while earlier vintages of wind and solar installations contributed nothing to frequency regulation or operating reserves, current vintages can, and in some countries network codes are being revised to require this functionality.[1] System operators have also modified their load forecasting to incorporate more detailed information on anticipated wind speeds, significantly improving unit commitment and dispatch decisions. Unintended Impacts Some of this adaptation has lagged, creating economic conflict and losses. For example, policymakers in many countries used out-of-market payments (e.g., feed-in tariffs, production tax credits) to incentivize investment in renewables. This approach has depressed electricity wholesale prices—even driving them negative in some hours of the day—with the uneconomic result of pushing a number of legacy nuclear plants to be retired prematurely. When existing nuclear plants are shut down, their generation is typically replaced by either natural gas and coal or a mix of variable renewables and fossil fuels. As a result, the carbon footprint of the electric grid inevitably increases, as observed following the closure of US nuclear plants (e.g., Crystal River in Florida [2009], Kewaunee in Wisconsin [2013], San Onofre in California [2013], Vermont Yankee in Vermont [2014], and Pilgrim in Massachusetts [2019]).[2] Policies governing wholesale market design and the electricity sector need to be updated for the new reality of a grid that accommodates both variable renewables and other low-carbon technologies to exploit the contributions of each to the decarbonized grid. New policies should allow existing nuclear plants to continue to operate, avoiding emission increases that set back the gains of other low-carbon sources (as seen for example in California and Germany). The Capacity Planning Problem All electricity systems, however organized, must find a solution to the capacity planning problem: What portfolio of technologies and power plants should be built to serve future load? One of the criteria applied is minimization of total system cost,[3] which includes both the up-front investment in capacity and the later expenses of operation, including fuel costs. Historically, because dispatchable thermal technologies have dominated most systems, portfolio optimization sorted technologies into categories such as baseload, load-following, and peaker (operating only at times of peak demand). The technologies were primarily distinguished by the trade-off between fixed costs (mostly capital costs, but for nuclear also fixed operating costs) and variable costs (primarily fuel). High fixed cost technologies can yield a low average cost of generation so long as they are operated with a high capacity factor. They are therefore chosen to serve the baseload portion of the load curve. This is the market niche in which nuclear has historically competed. Low fixed cost, high variable cost technologies can yield a lower average cost of generation when the capacity factor is low, so they are chosen to serve the peak load portion of the load curve. Combustion turbines are an ideal technology for this niche. The optimal mix of technologies is determined by their fixed and variable costs and by the load curve. Profile of Annual Renewable Resource Availability The introduction of renewables into the set of available technologies brings in a new factor: the profile of renewable resource availability through the hours of the year, with daily, seasonal, and synoptic variability. Expanding the use of renewables can add a high volume of generation to some hours of the year, but not others. Figure 1 Figure 1 illustrates the problem. The left side shows the hourly load in the six New England states in 2018; it appears as a dense, gently undulating blue band of hourly data points. The width of the band reflects the daily fluctuation in load: it expands in the summer months, and peaks there, too, with a secondary peak in the winter. The left side also shows, in red, a simulation of renewable generation from a hypothetical portfolio of wind and solar PV facilities. The size of the portfolio is chosen so that total renewable generation for the year equals total load for the year. For this thought experiment we did not analyze whether it is environmentally sustainable to deploy PV panels, wind turbines, and their associated transmission infrastructure on such a grand scale. The fluctuations in hourly generation reflect the varying insolation and wind across the hours. These are much larger in scale than the fluctuations in load. Although the simulated renewable generation matches load in the aggregate over the year, within any shorter interval there is a large mismatch. For this portfolio of renewable capacity to successfully serve load, it would have to be complemented with facilities that can store the surplus electricity in some hours and release it back in other hours. The right side of figure 1 shows in green the hourly charge and discharge (surplus or deficit) of a hypothetical lossless storage system, along with the total state of charge (storage; purple). While the storage is used to smooth the daily fluctuations in generation, the figure makes clear that there is a large seasonal cycle to the storage. The total capacity of the storage system must be nearly 14 TWh, which is enormous. Alternatives to Storage One alternative to a seasonal storage system is to enhance wind and solar capacity so that even in the low resource hours there will be sufficient generation to meet load. Doing so would mean that total renewable generation capacity would far exceed total load, and in some hours there would be very large curtailments. Figure 1 shows the limits of this option: there are quite a few hours when the total renewable resource is very low, so that an extremely large amount of capacity would be needed. Another alternative is to invest in extensive long-distance grid connections that diversify the variable generation assets accessible to load. The most viable alternative is to identify other low-carbon generation technologies for use whenever renewable generation is too low. These include nuclear, reservoir hydro, geothermal, hydrogen, and biofuels. A survey of studies on deep decarbonization pathways identifies this role of “firm” (i.e., reliable) low-carbon resources as critical (Jenkins et al. 2018a). This is a new way of thinking about constructing an optimal portfolio of generation technologies. It is no longer enough to focus on the load curve. It is now necessary to appreciate the interaction between the time profile of load and resource availability. Expanded investment in renewables adds a high volume of generation to some hours of the year, but not others. While the dramatic drop in costs has made wind and solar PV the economic choice for incremental investment, as penetration expands additional investments fall because these resources are not serving load in the most deficient hours. Other technologies are required to serve these hours or to store energy from low-carbon generation and deliver it to these hours. Table 1 Table 1 contrasts the old and new ways of thinking. In the new taxonomy (proposed by Sepulveda et al. 2018), nuclear competes among firm low-carbon resources complementing intermittent renewables. While nuclear has a place in both taxonomies, its place in the portfolio changes. The Opportunity for Nuclear Energy in a Decarbonized Electricity System We have examined nuclear’s new role in a decarbonized electricity system (Buongiorno et al. 2018). Projecting to 2050, we asked, What are least-cost mixes of generation technologies to serve loads in diverse regions while achieving targeted reductions in carbon intensity? We paid particular attention to how the accelerated growth of variable renewable technologies such as wind and solar PV alters the optimal portfolio mixes. Explanation of Our Model We applied a capacity expansion and dispatch model to the conditions in a variety of regions with different load and renewable resource patterns. We chose six regions, two in the United States (New England and Texas), two in China (Tianjin-Beijing-Tangshan and Zhejiang), and two in Europe (France and the United Kingdom). In this paper we focus on the US regions. We used the GenX model, a constrained optimization model that determines the least-cost mix of investments required to serve electricity demand in a future planning year (Jenkins and Sepulveda 2017). The optimization criterion is total system cost, which includes the capital expenditures to install the capacity as well as the subsequent operating expenditures. The model assumes that capacity is dispatched and operated to minimize the total system cost, subject to constraints. The constraints include the requirement that net generation equal load in each of the 8760 hours of a representative year, taking into account the availability of storage and demand response. Hourly renewable generation is constrained by installed capacity and by the hourly availability of the renewable resource. Generation units must operate within their technical constraints, such as minimum load, maximum ramping capacity, and so on. For example, we assume that nuclear plants have a minimum generation level of 50 percent and can ramp at a rate of 25 percent per hour, while gas turbines have a minimum generation level of 24 percent and can ramp at 100 percent per hour.[4] Finally, aggregate CO2 emissions must be within a specified constraint. GenX can be configured for different levels of detail. For example, it can incorporate opportunities for demand response as well as certain defined transmission constraints; the study discussed here did not include the former and treated each region as if it had no transmission constraints and no trade outside the region. GenX can also be parameterized to include existing capacity and to choose new investments, but the study discussed here focused on a greenfield mix for 2050—i.e., with no inherited capacity. The exception is hydro facilities, which were fixed at the existing level. The model requires inputs on the available technologies, their operating constraints, and capital and operating costs. The technologies included were utility-scale solar PV, on-shore wind, large-scale traditional nuclear reactors, natural gas with carbon capture and sequestration (CCS), coal with CCS, open-cycle gas turbine, combined-cycle gas turbine, coal, pumped-hydro storage, and battery storage. This is a limited set of technology options, but its range is broad in terms of key characteristics. Technologies compete among each other in important ways. The options included in the analysis and the cost inputs chosen shape the results. The full report details a number of scenario analyses performed to expand on the basic results. Total System Cost A portfolio optimization that looks at total system cost is essential, and far superior to comparisons of levelized cost of electricity (LCOE) numbers for competing technologies. While LCOEs can be useful summary benchmarks of the different cost inputs, comparing LCOEs across technologies implicitly assumes the technologies compete head-to-head to serve similar loads. As explained above, in reality, technologies are often best suited to serve certain portions of the load and ill suited to serve others, so that what may seem to be competing technologies are actually complementary. Our implementation of GenX addresses this issue and brings out both the opportunity for nuclear and the challenge of cost. Details on the cost assumptions are available in the MIT study report (Buongiorno et al. 2018), but some key figures are useful to mention here. The nominal or base case assumption for the overnight capital cost of nuclear in the two US regions for 2050 is $5500/kW in 2014 dollars (we also considered a case in which the capital cost for nuclear is decreased by 25 percent to $4100/kW; the study contains a number of sensitivity analyses on various parameters). A concise summary of these inputs is the LCOE of key technologies. Assuming the nuclear plant operates at a 90 percent capacity factor, the LCOE is just over $100/MWh. In contrast, assuming capacity factors for wind and solar of 34 percent and 25 percent, respectively, and approximating the available resource factors across our US regions, the LCOEs for wind and solar are $72/MWh and $52/MWh. So nuclear is an expensive alternative. Yet it may be a valuable part of a portfolio because of its capability to generate during hours when the renewable technologies are less available. Table 2 We calculated the optimal portfolio in each region under different assumptions about the level of decarbonization as measured by the carbon intensity of the system, starting at 500 gCO2/kWh and falling to 100, 50, 10, and finally 1 gCO2/kWh.[5] To benchmark these different levels, table 2 contrasts the recent historical level of carbon intensity of the electricity sector with the 2050 goal established for a scenario developed by the International Energy Agency (IEA 2017) to be consistent with a 2°C target for global warming. For example, the 2014 US level was 486 gCO2/kWh, while the 2050 goal is 11. For China it was 698 gCO2/kWh in 2014, and 24 in 2050. Figure 2 Figure 2 shows optimal builds of new nuclear capacity in New England depending on the level of decarbonization targeted and the assumed cost of a plant. The blue bars show the results for the nominal case assumptions. Given the high cost assumed for nuclear and the quality of renewable resources in New England, significant emission reductions are accomplished without nuclear: both the 100 and 50 gCO2/kWh carbon intensity targets can be theoretically achieved at lowest cost without any nuclear capacity. However, deep decarbonization brings nuclear into the optimal mix. Figure 2 shows that (i) for deep decarbonization (emission intensity levels of 10 to 1 gCO2/kWh) nuclear is an important element of the portfolio that minimizes system cost and (ii) lowering the cost of nuclear makes a dramatic difference to the scale of nuclear—although it is a potentially valuable player in a decarbonized grid, cost is a determining factor for its scale. The MIT report includes a number of scenario analyses with varying cost assumptions, including about the cost of alternative firm, low-carbon technologies; naturally, nuclear’s role in the cost-minimizing portfolio varies with these assumptions. While figure 2 emphasizes the impact of lower cost on nuclear’s role in a portfolio, the reverse is also true: if nuclear projects in the United States and Europe continue to have cost overruns (as recent projects have), then nuclear will not play an important role in a cost- or carbon-minimizing portfolio. Figure 3 Figure 3 reports the average generation cost at each decarbonization target and for different assumptions about nuclear. In all cases, costs increase as the target for carbon intensity becomes tighter. The orange bars show the impact on costs of excluding nuclear from the mix. The figure makes clear that excluding nuclear is very expensive in terms of climate change mitigation, and its inclusion is comparable with other energy options. The GenX analysis shows that the use of nuclear energy in regions and nations is regionally dependent, which one would assume given the variability of solar and wind as well as the costs to deploy the technologies. Conclusion Meeting the world’s energy needs while simultaneously reducing greenhouse gas emissions is an enormous challenge. Meeting this challenge in the electricity sector will require a new mix of generation assets. While a variety of low- or zero-carbon technologies can be used in various combinations, our analysis shows the potential contribution of nuclear as a firm low-carbon technology. It is time to transform thinking about energy production. Renewable and nuclear energies are complementary, not mutually exclusive. Existing nuclear power plants should be preserved and new ones designed and delivered. References Buongiorno J, Corradini M, Parsons J, Petti D. 2018. The Future of Nuclear Energy in a Carbon-Constrained World. Cambridge: Massachusetts Institute of Technology. Davis L, Hausman C. 2016. Market impacts of a nuclear power plant closure. American Economic Journal: Applied Economics 8(2):92–122. EIA [US Energy Information Administration]. 2019. International Energy Outlook 2019. Washington. Ela E, Gevorgian V, Fleming P, Zhang YC, Singh M, Muljadi E, Scholbrook A, Aho J, Buckspan A, Pao L, and 5 others. 2014. Active Power Controls from Wind Power: Bridging the Gaps (No. NREL/TP-5D00-60574). Golden CO: National Renewable Energy Lab. EPRI [Electric Power Research Institute]. 2014. Program on Technology Innovation: Approach to Transition Nuclear Power Plants to Flexible Power Operations 2014. Palo Alto CA. IEA [International Energy Agency]. 2017. Energy Technology Perspectives 2017. Paris: OECD/IEA. Jarvis S, Deschenes O, Jha A. 2019. The Private and External Costs of Germany’s Nuclear Phase-Out. Working Paper 26598. Cambridge MA: National Bureau of Economic Research. Jenkins JD, Sepulveda NA. 2017. Enhanced decision support for a changing electricity landscape: The GenX configurable electricity resource capacity expansion model. Massachusetts Institute of Technology Energy Initiative Working Paper. Cambridge MA. Jenkins JD, Luke M, Thernstrom S. 2018a. Getting to zero carbon emissions in the electric power sector. Joule 2(12):2498–510. Jenkins JD, Zhou Z, Ponciroli R, Vilim RB, Ganda F, de Sisternes F, Botterud A. 2018b. The benefits of nuclear flexibility in power system operations with renewable energy. Applied Energy 222:872–84. Keppler JH, Cometto M. 2012. Nuclear Energy and Renewables: System Effects in Low-carbon Electricity Systems. Paris: OECD/NEA. Miller N, Lew D, Piwko R. 2017. Technology capabilities for fast frequency response. GE Energy Consulting Technical Report 3:10–60. Neidell MJ, Uchida S, Veronesi M. 2019. Be Cautious with the Precautionary Principle: Evidence from Fukushima Daiichi Nuclear Accident. Working Paper 26395. Cambridge MA: National Bureau of Economic Research, Ponciroli R, Wang Y, Zhou Z, Botterud A, Jenkins JD, Vilim RB, Ganda F. 2017. Profitability evaluation of load-following nuclear units with physics-induced operational constraints. Nuclear Technology 200(3):189–207. Sepulveda NA, Jenkins JD, de Sisternes FJ, Lester RK. 2018. The role of firm low-carbon electricity resources in deep decarbonization of power generation. Joule 2(11):1–18. Varma RK, Akbari M. 2019. Simultaneous fast frequency control and power oscillation damping by utilizing PV solar system as PV-STATCOM. IEEE Transactions on Sustainable Energy 11(1):415–25. Wu Z, Gao W, Gao T, Yan W, Zhang H, Yan S, Wang X. 2018. State-of-the-art review on frequency response of wind power plants in power systems. Journal of Modern Power Systems and Clean Energy 6(1):1–16. [1] A report prepared for the Australian Energy Market Operator gives a good feel for the evolution in this area (Miller et al. 2017). See also Ela et al. (2014), Varma and Akbari (2019), and Wu et al. (2018). [2] For documentation of the impact of nuclear plant closures on CO2 emissions, see Davis and Hausman (2016) re San Onofre, Neidell et al. (2019) re Japan’s closures immediately after Fukushima, and Jarvis et al. (2019) re Germany’s policy decision to close its nuclear plants. [3] Cost minimization is a crucial but admittedly narrow focus. The energy mix should be and typically is determined by consideration of broader economic impact (jobs, taxes, business opportunities), local environmental impact (air quality, land use), fuel supply security and diversification, resilience of the energy infrastructure, geopolitical relationships, etc. [4] Contrary to popular belief, the output of many large nuclear power plants in Europe and the US is routinely adjusted according to system requirements. Many provide frequency regulation service, others operate in a load-following mode at daily and weekly scales, and some adjust to seasonal needs. See EPRI (2014), Jenkins et al. (2018b), Keppler and Cometto (2012), and Ponciroli et al. (2017). [5] Our modeling measures and constrains only direct emissions. All technologies have so-called indirect emissions attributable to the infrastructure and the supply chain. The direct emissions from fossil fuels are outsized relative to the indirect emissions from most technologies and so have been the focus of policymakers. Moreover, if a policy addressing direct emissions is broad enough, encompassing most sectors, then the indirect emissions from one sector will be captured as direct emissions in another. Eventually, as large sources of direct emissions decline, the relative variation in indirect emissions will gain attention. About the Author:Karen Dawson is a consultant with Bain and Company. Michael Corradini (NAE) is emeritus professor of nuclear engineering and past director of the Wisconsin Energy Institute at the University of Wisconsin. John Parsons is a senior lecturer at the Sloan School of Management, and Jacopo Buongiorno is the TEPCO Professor of Nuclear Science and Engineering, both at the Massachusetts Institute of Technology.