Can Nuclear Energy Significantly Reduce the Cost of Electricity in Australia?

The case for nuclear energy based solely on electricity cost is marginal

By Matthew Rogozinski

Consider the following scenario: It's 2025, and we must decide whether to incorporate nuclear energy into the National Electricity Market (NEM).

The AEMO's Integrated System Plan 2024 (ISP) for the NEM does not consider nuclear energy; it envisages the lowest-cost deployment of solar PV and onshore and offshore fixed wind, firmed up with flexible gas and integrated into the grid to ensure its reliability, along with the 'behind-the-meter' generation and storage. Should we continue with this plan or do less of the above and incorporate nuclear energy into the NEM? In particular, which option is cheaper?

Our analysis shows that the case for nuclear energy based on electricity cost alone is marginal at best. However, the case is not weak enough to reject nuclear energy based on its cost without a deeper investigation. AEMO should be directed to include nuclear energy in its analysis and present the results before releasing the next ISP in 2026.

To decide whether to incorporate nuclear energy into the NEM, we consider utility-scale generation and storage, as consumers and businesses will determine how much solar and storage they install 'behind the meter'. We only consider the capacity to be deployed and integrated into the grid from 2025 onwards. In other words, capacity and its integration into the grid until 2025 are a sunk cost irrelevant to the decision we are about to make.

To make the decision, we calculate the levelised cost of electricity (LCOE) based on the ISP's incremental, integrated renewables firmed up with gas and compare it to the LCOE of large-scale nuclear-reactor-generated electricity becoming available in 2035. Integration of renewables into the grid comprises storage—pumped hydro and batteries—new transmission lines and connections, and grid strength remediation. We assume gas is deployed with carbon capture and storage (CCS)—an expensive option—to make it comparable to nuclear energy's low emissions.

The results of our analysis are in the chart below, which shows the LCOE range between high and low capacity factors for the technologies deployed. The upper/lower range LCOE corresponds to each technology's lower/upper capacity factor. The capacity factors for all technologies used in our analysis are from the CSIRO GenCost 2023-24 report, including the nuclear energy capacity factors ranging from 53% to 89%. In addition, we present the nuclear LCOE for capacity factors ranging from 70% to 94%. These nuclear reactor capacity factors represent the experience in France and the US. Nuclear energy represents about 70% of the grid in France and is generated in the load-following mode (matching the demand). In the US, nuclear energy contributes about 20% of electricity and is used as a baseload, thus operating with capacity factors up to 94%. At small nuclear capacity, say, 6 GW, the reactors would be used as baseload in the NEM, so high capacity factors should be feasible.

Nuclear energy generated by reactors operating with capacity factors between 53% to 89% is not cost-competitive with renewable energy integrated and firmed up with gas (when we compare LCOEs for upper and lower capacity factors). However, with capacity factors ranging from 70% to 94%, nuclear energy is cost-competitive with offshore wind and gas with CCS, the most expensive utility energy sources in the ISP version of the NEM.

Furthermore, some renewables' integration costs, which range from $26 to $43 per MWh, would be avoided with the offshore wind capacity reduction due to the incorporation of nuclear energy.

So, based on the cost alone, nuclear energy could replace offshore wind and possibly gas in the NEM. However, since nuclear energy costs are comparable but not meaningfully lower, its inclusion at a small capacity would only slightly lower the NEM's incremental generation and storage cost. In other words, the case for nuclear power based solely on electricity cost is marginal.

We used a 60-year timeframe for our analysis. While nuclear reactors have been operating in the US for an average of 40 years, the Nuclear Regulatory Commission has licensed 88 of the US 92 reactors to operate for another 20 years, and 15 are applying for a second 20-year extension. Our conclusions are the same if we assume an 80-year timeframe. The nuclear cost is reduced by only 2% when extending the lifetime by a third from 60 to 80 years.

The results are highly sensitive to the capital investment. When the capital investment is reduced by a third, the nuclear LCOE is reduced by nearly a third. However, the CSIRO 2030 large-scale nuclear reactor capital expenditure estimate used in our analysis is already below the US Department of Energy's most optimistic repeat-deployment cost projection (the 'advanced' scenario), using the OECD's 2023 purchasing power parity exchange rate. This comparison cautions against assuming a lower nuclear capital investment.

Finally, we mention some other assumptions in our analysis. We assumed all capacity decommissioning envisaged in the ISP is equipment installed before 2025 and did not incorporate it into the analysis. However, renewable and gas generation and storage equipment installed starting from 2025 will be decommissioned and replaced, which we accounted for in our analysis as follows: solar panels every 30 years; wind turbines every 25 years; gas turbines every 25 years; batteries every 15 years; pumped hydro every 50 years. All current and projected unit costs, construction periods and capacity factors are per the CSIRO GenCost 2023-24 report's NZE Post 2050 scenario corresponding to the ISP's Step Change scenario. The discount rate is 7% per the ISP's central assumption. We base our analysis on capacity factors rather than observed renewable energy traces. We did not quantify the cost of energy curtailment. The results are not adjusted for risk.