Quality In, Quality Out

Yes, the Federal Minister for Climate Change and Energy tweeted some unparalleled cringe about nuclear power, but let’s do something positive with it.

Original tweet

1. Cost Methodology

To reach an uncharitably high pricetag of $5 billion per power plant, the minister multiplied the gross nameplate output of the small modular reactor design specified in work by Nuclear For Climate, i.e. 300 megawatts (or 300,000 kW), with the overnight capital cost number for nuclear supplied by the CSIRO since 2017: $16,000 per kW. This gave $4.8 billion, which was rounded up.

The SMR design is GE Hitachi’s BWRX-300, a remarkably simplified small boiling water reactor that is on track to first be constructed by 2028 in Ontario, Canada. It is a “Generation III+” design.

Originally, the CSIRO’s capital cost number was applied to an imaginary Generation IV reactor (page 89) with erroneous referencing and no apparent consultation with relevant experts. Furthermore, it is overnight capital cost (OCC), meaning that interest and other project costs are not captured. Multiplying it by a reactor’s capacity, as the minister has done, provides next to no information.

This is the deficient standard of analysis set by the federal minister. We’ll do substantially better in the following comparison.

2. Small Modular Reactors in the Australian Context

All else being equal, what is a fair and reasonable set of estimates for the total project costs of small modular reactors in Australia?

The answer requires defensible values for (1) overnight capital cost, (2) total project cost including financing (as defined here, page 38), and (3) a way to anticipate the learning-by-doing cost decline as First-of-a-kind progresses to Nth-of-a-kind.

The 2nd Edition of Ben Heard’s Small Modular Reactors in the Australian Context quotes GE Hitachi’s target overnight capital cost for the 280 MW(net) BWRX-300, converted to Australian dollars with a conservative 25% additional loading for Australia-specific owner’s contingency costs. This equals $4,000 per kilowatt installed for Nth-of-a-kind. We will use it as-is to be conservative.

The real First-of-a-kind OCC value won’t be known until OPG decides to make it public. This project is expected to expand to multiple units, followed by projects for the Tennessee Valley Authority, industrial users in Poland, and for further Canadian provinces. To remain conservative and defensible we’ll calculate the total project cost for Australia as First-of-a-kind.

The second and third factors are outlined in Unlocking Reductions in the Construction Costs of Nuclear published in 2020 by the OECD-NEA. This document draws on considerable international project experience, and is so relatively exhaustive that it should be compulsory for any given analysis of nuclear project costs.

These are the the relevant charted material:

From the report: “Capitalised financial costs can make up more than 20% of total investment costs per kWe, especially for projects that take more than five years to construct and have a cost of capital higher than 7% — the case for most nuclear projects under construction in OECD countries.”

From the report: “Figure 14 provides a representative breakdown of investment costs per kWe of nuclear power plants in OECD countries. [Engineering, Procurement and Capital] claims more than 50% of the total investment costs, followed by [Interest paid to investors During Construction] (~25%) and owner costs and contingencies (15% each).”

From the report: “[First-of-a-kind] costs can be defined as “non-recurring” and usually include design and testing, licensing and certification, and supply chain training and qualification, which are essentially of an indirect nature. Nonrecurring costs can add approximately 30–35% to the OCC of a first reactor…”

These allow us to coursely estimate recurring and non-recurring investment costs.

3. Calculations

We take the Nth-of-a-kind overnight capital cost plus Australian owner’s costs and contingencies, in Australian dollars, increase it according to Figure 15 (dark blue bars) and add the Non-recurring costs (light blue bars).

Target NOAK OCC × FOAK/NOAK Recurring cost (10/7) + Non-recurring costs (35%) = Australian FOAK OCC

$4,000 × 10/7 + $2,000 = $7,714 per kW-installed

Then we add the Interest paid to investors During Construction (IDC), according to Figure 11 and its explanatory text, before multiplying it by the SMR’s net capacity.

Total FOAK project cost for the first BWRX-300 power plant:
at nominal 25% IDC: $7,714 + $1,928 × 280,000 kW = $2.7 billion
at high 40% IDC: $7,714 + $3,085 × 280,000 kW = $3 billion

The nominal IDC result can be justified on the basis of a social discount rate (reduced cost of capital) and substantial foreign orders for BWRX-300 units prior to Australia’s project commitment (in the 2030s).

Second-of-a-kind OCC and total project cost can be expected to decline. This is illustrated in Figure 15 as a 5/8 reduction in investment cost and is broadly consistent with independent analysis in the ETI Nuclear Cost Drivers summary report.

Australian FOAK OCC × 5/8 = Australian 2OAK OCC

$7,714— $2,893 = $4,821 per kW-installed

Total 2OAK project cost at 25% IDC: $4,821 + 1,205 × 280,000 kW = $1.7 billion

Conservatively, Nth-of-a-kind SMR builds would reach OCCs of $4,000 and total project costs of $1.4 billion.

4. Comparing to Semi-equivalent, Government-Supported Solar Capacity

Australia’s first utility scale photovoltaic power station was constructed in Nyngan, NSW in 2015. Output over each financial year is recorded by the Clean Energy Regulator and indicates a healthy annual average of 235.6 gigawatt hours. Nyngan cost $290 million ($312 million in 2020 AUD), a little over half of which was entirely supplied by federal and state grants.

Most recent year’s worth of Nyngan PV daily generation, courtesy of OpenNEM. For comparison, 24 hours × 280 MW = 6,720 MWh.

At an average 90% capacity factor, the BWRX-300 SMR is expected to generate 2,365 gigawatt hours per year, with a design life of sixty years. Briefly stooping to the Minister for Climate Change and Energy’s level of mathematical rigour, we would need ten Nyngan solar farms to match just the bulk energy of one SMR, at a cost of $2.9 billion. If we naively compare this to the $2.7 billion SMR FOAK cost, the minister’s rhetoric holds up quite poorly.

We cannot say for how long large-scale solar might have been delayed in Australia without direct government support, which had the effect of mitigating project risk and depressing the cost of capital, i.e. a social discount rate.

But, naturally, photovoltaic power plant costs have fallen since the first government-supported examples, e.g. the new Moura 110 MW project in Queensland cost $120 million last year. Such is the benefit of consistent learning-by-doing for energy projects which enjoy minimal opposition, and it is why all else needs to be equal.

5. What About Time?

We are not even close to being on track to achieve near-term climate goals in our electrical sector.

As the Australian Energy Market Operator stated in 2018, “there are extents of [variable renewable energy] penetration that have not been explored, and for which there is no experience globally” and as such, any and all confidence in unprecedented combined solar and wind shares relies entirely on models. However desirable, annual renewable energy penetrations approaching 80% or above may not be realistically achievable.

Live 24/7 CO₂ emissions of electricity consumption

With the support of a provincial utility, a national government and an existing nuclear sector workforce, the first BWRX-300 can reasonably be expected to start operating in five years time, in a region already leading in consistently low-emission electricity. Even if Australia wisely removed the unjustified prohibition on nuclear energy now and embarked on instituting the necessary regulatory structure — even if merely as a “Plan B” — we still need to get in line.

Oscar Archer holds a PhD in chemistry and has been analysing energy issues for over 15 years, focusing on nuclear technology for nine, with a background in manufacturing and QA. He helps out at RePlanet Australia. Find him @OskaArcher on Twitter.