Read this for the context.
The first set of scenarios looks at the possible build out of Gen II+/III/III+ thermal reactors (i.e., current and advanced water-moderated reactors: PWRs, BWRs, HWRs etc.), over a 50 year time frame (2011 to 2060). The focus of this exercise is not to predict which reactor type(s) will predominate. In the next 1-2 decades, I suspect (given current and announced installation schedules) that it will be dominated by a mix of monolithic designs, such as the AP1000, APR-1400, CPR-1000, EPR, ABWR, PHWR, VVER-1000, as well as some small modular reactors like the mPower and NuScale.
The starting installed capacity in 2010 is set at 376 GW (all power rates are expressed as electrical rather than thermal output). The projection model is based on 5-year blocks (2011-2015, 2016-2020, etc.), for which a growth rate (multiplier) is specified, through to 2056-2060. Four scenarios are considered:
1. NCOL: WNA Nuclear Century Outlook (NCO) Low (anchoring to 602 GW in 2030 and 1140 GW in 2060)
2. NCOH: WNA High scenario (1350 GW in 2030, 3688 GW in 2060)
3. TR1: A mid-growth scenario that tracks between NCO Low and High, but which peaks at around 2050 and slowly declines thereafter
4. TR2: A high-growth scenario that is identical to NCO High through to 2030, after which the relative growth rate slows only gradually (absolute number of GW per year continues to increase).
The following table summarises the four scenarios:
Here is a plot of the four scenarios, in terms of GW of installed capacity, for the period 2011 to 2030 (recall that NCOH and TR2 are identical for this period):
Extending the projections for another 30 years, through to 2060, we get this:
As you can see, the TR2 scenario reaches nearly 10,000 GW (10 TW) by 2060, so is obviously highly ambitious. But it matches the probable global electricity need, justified here. (The charts are not completely smooth because of the assumption of discrete 5-year blocks, rather than some continuous time function for growth rates.)
Here is a tabulated form:
The remaining energy gaps will be filled by other technologies, including existing coal and gas, fossil fuel plants with CCS, hydro, geothermal and new renewables. I don’t propose to speculate on what their relative proportions might be, although I expect hydro to maintain about its current level and coal to play the largest role through to 2030.
Annual fuel use is set at 170 tU (tonnes of uranium metal) per gigawatt year (GWy). Sufficient depleted uranium (DU) is assumed to always be available for initial loadings. The annual production of plutonium (Pu) in the spend fuel is set at 0.25 t/GWy. If used in mixed oxide fuel, is is assumed to reduce uranium usage by 15%. For justifications of these parameters, read this.
Capital cost in 2011 is set at $4 billion/GW, and projected to decline linearly to $2 billion/GW by 2031, and fixed thereafter.
This analysis is run on an Excel spreadsheet (download Excel 2010 or Excel 1997-2003 compatible versions). I will continue to update these as I proceed with the modelling. You can use this to add other scenarios, or tinker with other input parameters, create extra plots, etc.
Scenario outputs: The following table summarises the key outcomes of the four scenarios (click it for an enlarged version; the units of columns 2-5 are in tonnes, c6 is GW installed):
Mined uranium ranges from a little greater than today identified reserves of 5.5 Mt (at <$130/kg U), through to over 30 Mt cumulative use by 2060. Between 9 and 45 kt of Pu will exist in the used thermal reactor fuel by 2060 — this has significant implications for later projections that involve IFR or LFTR build out schedules. The maximum number of 1-GW plants that would be built in a given year ranges from a low of 23 for the NCOL through to a little over 1 GW a day for TR2 (this peaks in 2040, and has declined to 286 by 2060; the 2030 rate was 130 GW/yr).
Capital costs for the NCOL average at $35 billion/yr for 2011-2030, and $37 B/yr for 2031-2060. For TR2, which has reached a vast capacity of 9,835 GW by 2060, the average cost is $122 B/yr to 2030, and $570 B/yr.
To put the plausibility of these scenarios under the spotlight, I’ll reiterate what I said here. (These are the real-world facts as they stand).
In the short term, a total of 26 gigawatts (GW) of new nuclear plant will start operation in 13 different countries in the 2010 — 2012 period – that’s within the next 3 years (average reactor size is 880 MW). Nuclear power is being most actively pursued today in China (25 reactors currently under construction), India (4), South Korea (6) and Russia (8), and in terms of forward projections through to 2020, China plans to expand its nuclear generation capacity to 70 GW (up from 8.6 GW in 2010), South Korea to 27.3 GW (up from 17.7 GW), and Russia from 43.3 GW (up from 23.2 GW). Looking further ahead, India’s stated goal is 63 GW by 2032 and 500 GW by 2060, whilst China’s 2030 target is 200 GW, with at least 750 GW by 2050.
The two leading reactor designs now being built in China are the indigenous CPR-1000 and the Westinghouse AP-1000. Reported capital costs are in the range of $US 1,296 to $1,790/kW. Korea has focused attention on its APR-1400 design, with domestic overnight costs of $2,333/kW. A recent contract for $20.4 billion has been signed with Korean consortium KEPCO to build four APR-1400 reactors in the United Arab Emirates, at a turnkey cost of $3,643/kW (i.e., $3.6 billion per GW). This price is notable considering that it is offered under near-FOAK conditions, because these will be the UAE’s first nuclear plants.
In the next few SNE2060 posts, I will examine some of these assumptions and constraints in a little more detail, before moving on to Gen IV possibilities. Particularly: (i) Are the build rates implied here plausible?; (ii) Is there enough uranium to fuel the sort of growth in thermal reactors and what will these demand curves do to fuel price? and (iii) What are the implications for repositories for the spent fuel? and (iv) How sensitive are these results to the various assumptions (including the results of some sensitivity analysis).