Why I changed my mind on nuclear

I've been thinking about decarbonising for a long time; I first got involved in the mid 90's. I was working on gas combined heat and power then and it was great way to reduce emissions; it isn't today. And therein lies the problem with nuclear; the world moves on.

In 1993, the 6MW Royd Moor Wind Farm was commissioned fairly locally to me - it's not the oldest in the country, but it's somewhere in the earliest 10 (still running 31 year later and due to decommission in 2027). Solar was non-existent then. Nuclear seemed like the only option that could get us anywhere close to a low carbon future. By the early 2010's wind was just going offshore at around £150/MWh; nuclear still looked cheap by comparison, especially considering the intermittency issues and the cost of batteries at that time, but please don't assume at this point I'm going say batteries are the answer!

By the mid 2010's, the astonishing reductions in the cost of renewables were making me question whether nuclear was the right solution, especially when I could see the deal struck for Hinkley C at £92.50/MWh (2012 prices - I'll use this basis throughout because that's how UK CFDs are priced - see the 90% capacity factor end of the orange line on the chart below). But there was still the problem of how to firm renewables to match demand.

Fast forward to today and Hinkley, like almost every other nuclear build, is way over budget and behind schedule. In 2016 when the project was given the go ahead, the cost basis was £18bn, the latest indicators are that it could be as much as £46bn. Let's be optimistic and assume it might come it at £36bn, just 100% over budget; the £18bn overspend would be enough to buy around 9GW of offshore wind generating about 35TWh per year - about 40% more than Hinkley will produce in a year; and that's just from the overspend. It would leave us with a handy £18bn to spend on firming that wind.

What do the numbers look like?

Proponents of nuclear are very quick to point out that LCOEs of individual technologies are not very useful in a world dominated by renewables and I concur; you have to consider the system cost which includes the cost of providing electricity when the wind isn't blowing. But let's not forget, that all the costs touted for nuclear assume very high capacity factors and those are not achievable either if nuclear is the dominant generation source; it will need to turn down when demand isn't there. Using some numbers from the recent Future Energy Scenarios, I estimate that the 'demand side' load factor in a net zero world might be around 65% - i.e. the ratio of peak winter demand (average cold spell) to the annual average demand. Using average cold spell here is probably reasonable because regardless of whether we use a nuclear solution or a firmed renewables solution, we will want to take advantage of short duration (battery) storage and some demand flexiblity to keep the load factor of nuclear high and the installed renewables back-up low. So if we have a future system operating with solely nuclear (+batteries), that nuclear plant would have to operate at a capacity factor of about 65%. I estimate that would push the LCOE of a nuclear solution built at the original estimated Hinkley C cost up to £135/MWh (2012 figures, see chart - orange line). If we assume a more realistic cost outcome, but delivering the same return on investment, then the LCOE at 65% capacity factor is more like £200/MWh (dotted orange line on chart). In practice there would be additional engineering costs to give it load-following capability, but let's be as optimistic as we can!

So for a renewables system to be cheaper than nuclear, we need to aim for a firmed LCOE of less than £135/MWh (being optimistic about nuclear costs). In the chart, I've shown the cost of offshore wind based on a 40% capacity factor and roughly today's prices (about £70/MWh in 2012 GBP). I've used 40% capacity factor as it's conservative given future expectations (up to 60%) and thus reflects that we might still need to turn wind off sometimes during very high generation periods. Let's assume that we can ignore the cost of batteries for intra-day, short duration, balancing as we'll want those for nuclear too. So we are trying to manage multi-day low wind conditions. There is probably scope for innovation in this area, but I'm going to use something we know we could do - hydrogen to power (H2P). We can build electrolysers, we have already stored hydrogen in salt caverns for over 50 years and turbine manufacturers have demonstrated GTs running on 100% hydrogen (albiet small ones so far due to the lack of hydrogen).

The chart shows the LCOE of H2P at different capacity factors in dark blue. It's expensive, mainly because making hydrogen is expensive and it's not very efficient. I've assumed that we have to build dedicated offshore wind (with electrolysers running at 40% capacity factor and 70% efficiency to allow for compression) to make the hydrogen, which comes in at about £150/MWh (£200/MWh in today's money, or £6/kg) and running through a power plant at 45% gross efficiency (remembering that hydrogen power plants will be lower efficiency than natural gas on a gross basis). I've also allowed £2.52/MWh for storage throughput.

A key question is: "How much of the time will renewables not meet the demand?" You can't answer that question with a simple chart, but when I run more time-granular analysis with an appropriate mix of wind and solar I get about 3-10% depending on various assumptions (including how much base load renewables or nuclear is included and how much over-build is included). This is not dissimilar to other whole system model analysis. The Royal Society suggest it might be around 20% in their "Large-scale Electricity Storage" report, but that includes for short periods of time where batteries and demand management (mainly from EVs avoiding charging) could provide the required capacity at little or no additional cost. It's notable that you do need the right mix of wind and solar to get the optimal solution; the great thing about wind is that it generates 2-2.5 times as much output during winter months, when we need it most, compared to summer months, so for the UK we need a wind-dominated mix. Let's suggest a range of 5-15% for our H2P capacity factor - the darker blue shaded bar on the chart. You can see that in this range, H2P is very expensive - from around £500/MWh at 5% to £390/MWh at 15%. But this is only needed for short periods of time; when we estimate a weighted average cost of electricity over the year it ranges from about £90/MWh to £118/MWh (light blue shaded area). That's less than our 'nuclear dominated' option using original Hinkley C pricing and operating at the required 65% capacity factor. It's a lot less than what we might estimate the true LCOE of Hinckley to be - about half in fact.

The only nuclear option that appears as if it could work is if we adopt the Rolls Royce's quoted SMR LCOE of around £65/MWh (it's hard to track this number down, but it's been slowly creeping up over time). From my chart that could be more cost effective than renewables + H2P, but its worth bearing in mind that I've been very conservative in my numbers - for example, I'm assuming all renewables are in the form of offshore wind, add in cheaper solar and onshore wind and my firmed renewables cost will fall, similarly my H2P figure could also fall if you believe the hydrogen proponents. Those things make SMRs look difficult economically. And what about that SMR cost? The world nuclear association says that the nuclear island in a power plant is about 28% of the total cost (that's remarkably similar to many other projects where the 'big bit' comes in at around 30%). The idea of an SMR is to make the nuclear island modular and reduce its cost, but even if you reduce it by 50% and assume that the balance of plant remains unchanged (in practice it will go up because vertical economies of scale are being lost by going smaller), the total cost would reduce by only ~15%. It's hard to see how that will lead to a LCOE which is less than half of where a large scale nuclear reactor currently sits.

But what about ancillary services and inertia, will be the nuclear lobby retort. Well we can provide virtual inertia from those batteries (with grid forming inverters) that we need for both renewables and nuclear. We can also provide all the other grid services we need from batteries and inverter connected renewables. If we want some real inertia, then we can build synchronous condensors/flywheels or, better still use clutched CCGT plant and let the H2P generators we've already paid for provide those services when the wind is blowing, helping to minimise its LCOE. There will also be an argument about additional transmission capacity needs; but bear in mind that UK offshore wind CFDs include the cost of transmission to a suitable connection point, whereas the cost of new Hinkley transmission is incurred through National Grid's regulated asset base. It's also the case that we are going to have to build more transmission and distribution capacity as electrifcation of heat and transport demand proceeds, this increased demand also means more MWh through the system and DESNZ suggest that the final unit cost of transmission and distribution may actually be lower than today.

History tells us that the cost of nuclear keeps going up whilst the cost of renewables (and storage) keeps coming down - there is no sign of that changing at present. It seems to me that nuclear is not needed to deliver a low cost, zero-carbon energy system, nor will it deliver a lower cost energy system (the Royal Society report I flagged early says the same). Politicians may have other reasons why they want to maintain civil nuclear capacity, but I don't think it can be justified from a pure energy system perspective.