Guest post by Energy Matters’ commentators Alex Terrell and Andy Dawson. In part 2 of their trilogy, Alex and Andy examine how the UK 2050 electricity demand may be met by a nuclear dominated supply model. It requires 85 GW of nuclear capacity in the UK. The model is founded on existing technology and existing UK nuclear sites. But as the decades pass goes on to include new UK nuclear sites previously occupied by coal fired power stations and clusters of small modular reactors (SMRS) that have yet to be built, licensed and tested. It concludes by introducing the concept of nuclear islands built in very shallow water off the English coast.
The previous article on 2050 electricity demand provided a scenario where the average electricity demand was approximately 72GW, but peak demand on exceptionally cold days could reach 121GW.
This article shows how this could be substantially fulfilled with nuclear power, relying on some gas (or biofuel) to provide peak power when required. A number of different scenarios are explored, with a preferred scenario of 85GW of available nuclear capacity.
The aim with this scenario will be to have 85GW of nuclear capacity available throughout the winter period. In practice, this will probably mean having 87GW of total nuclear capacity, with scheduled outages (for maintenance and refuelling) timed for the summer months.
The thermal capacity (mostly gas, but it could also contain some oil or biofuels) is used for peaking, especially in the winter months. The current UK electricity supply strategy already envisions 37GW of gas capacity. In this high-nuclear scenario, they only run with a utilisation of 6.5%.
Our next article, number three in the series, will explore how demand could be provided with a mix of wind and solar power, supplemented by gas and electricity storage. In reality, it is likely that a mix of nuclear, wind, solar, tidal, import cables and gas will be used, but the extreme scenarios provide useful reference points.
A Nuclear Capacity Model
The demand target
The previous article modelled the following electricity demand distribution curve for 2050.
Figure 1 – Frequency distribution of electricity demand in 2050. Whilst typical demand is around 60GW, extreme weather can push up demand (averaged over the course of the day) to 120GW.
Providing a system what can produce an average of 72GW, but up to 121GW (averaged over the day) on the coldest days is not straightforward. A demand of over 120GW might occur one day every three to ten years, but it still needs to be dealt with, and in a way which minimises the capital expense of nuclear plant (or renewables, as discussed in the next article), whilst minimising the use of fossil fuels.
Whilst current nuclear power can be ramped down, it is not currently economic to do so to any great extent, there may be future developments that could allow increased flexibility:
- Generation 3+ reactors such as the EPR can be ramped down to 60% of nominal output. In theory, doing so could extend the life of the reactor, but this is determined more by regulatory constraints than actual wear and tear. In addition, the cost of capital means extending the reactor life (decades in the future) is not such an advantage.
- Molten salt reactors may have a much lower capital cost, and could therefore economically be turned down or off. Designs based on graphite moderators (e.g. ThorCon, Terrestrial) have cores that are swapped out after a certain number of years. Reducing the power of these reactors will extend the core life and reduce the need for core swaps, making this solution more economic.
- Molten salt reactors will produce heat at a high enough temperature for steam methane reforming. The resultant CO2 could then be stored, and the hydrogen used to supply energy in the winter (either for direct heating, or by electricity generation in gas turbines or fuel cells). However, spare capacity would be in the summer, and demand for hydrogen will be highest in winter, and storing large quantities of hydrogen is not straight forward.
- High Temperature reactors (such as the pebble bed design being built in China)
may be able to switch from the production of electricity to the thermal production of hydrogen from water, whenever demand is low. This is a highly attractive solution to powering aviation, haulage and shipping (perhaps through synfuel production), but not without complications for energy storage. As above, spare capacity would be in summer, demand for hydrogen in the winter.
- High temperature reactors, including molten salt reactors, can store energy in the form of hot salts. The developers of the Moltex Reactor have proposed storing several hours worth of thermal output, which can be used to vary the electrical output. They have suggested this as a way of working with renewables, to counter their intermittency. However, this is not a solution for seasonal storage.
Given the uncertainty of the above technologies, it is prudent to assume – or at least to base the analysis on the assumption – that we want to run nuclear at a high capacity factor, and avoid building more capacity than is required.
Nuclear and gas solutions
The clear implication of the above analysis is that there’s a major incompatibility between the concept of an all-nuclear grid and the ability to supply the more extreme ends of the probability distribution for demand. However, the effects of this are not as severe as might immediately be assumed; carbon output is not particularly time-sensitive in that our key parameter is annual output, and hence short periods of what might appear to be high levels of fossil fuel output have little effect.
Taking the weather related demand model from the last 20 years of temperature records allows us to plot nuclear and gas plant utilisation on a daily basis, and to tabulate this into a capacity model.
Figure 2 – A nuclear capacity model. At low levels of nuclear penetration, the reactors work at their maximum available capacity (refuelling would result in extra gas being used). Only after about 50GW of capacity, does nuclear supply start to be curtailed (in the summer). An “available” capacity of 85GW nuclear would supply 97% of the electricity, and requires about 35GW of thermal power (gas, diesel, biofuel or hydrogen) capacity to supply the remaining 3%.
Figure 2 above shows this effect. With a nuclear capacity of 80 – 85GW average carbon intensity is of the order of 25-30g/kWh – well below the 50g/kWh threshold regarded as a “stretch target” by the Climate Change Committee.
Note that we’ve not attempted to model any effect from short term interventions available from the use of pumped storage, other than intra-day demand levelling. We anticipate a significant amount of pumped storage being available “on system” as it will have been key to a transition to a low carbon grid at current demand levels, before demand management becomes fully developed and delivers the diurnal stability which we have identified for this 2050 scenario.
We’re therefore led to the following preferred scenario:
Table 1 – A summary of the 2050 Nuclear supply scenario.
This compares with emissions from the energy supply sector of 153 Mt in 2014, which excludes emissions from nuclear, wind and solar. The carbon intensity of these sources is taken from the IPCC mid-range estimates (and is of course highly contentious).
The 85GW of nuclear capacity is the targeted winter availability. In effect this means having an additional 2GW of capacity to account for unscheduled down time in the winter (based on current US nuclear power plants “ forced outage rate” of between 1.1% and 3%.
The gas usage could be replaced by bio fuels if available. At the “tail –end” of the supply model, we could use diesel generators. These might not be used at all in a typical year, only coming into use in extreme circumstances – funded by a standby capacity payment. In this case, their high emissions do not materially impact CO2 emissions.
It should also be noted that the current Government / National Grid strategy expects a requirement of approximately 37GW of gas capacity by 2030. This would then fulfil the “peaking” capacity requirement for 2050.
Utilisations and maintenance
The next question we must consider is the sustainability of this level of supply against demand over the year – especially given that for a considerable proportion of the year demand is below that 85 GW.
In fact, the apparent mismatch is not problematic. A large proportion of demand will be fulfilled by Pressurised and Boiling Water Reactors – even if other technologies come on stream in the 2030s. These reactors do not “on-load refuel”; that is, they are periodically taken off supply, depressurised and a substantial proportion of the fuel load removed and exchanged (it’s also usual to “shuffle” the fuel that remains to optimise fuel utilisation). This would typically involve a 4-6 week shutdown occurring every 15-18 months, although there are technical options available to extend this to 18-24 month intervals. It is also usual to utilise these outages for overhaul and maintenance on those parts of the plant which are not accessible during normal operations. It leaves the majority of the world’s LWR fleet operating at a NET capacity factor of around 85-90%, including the effects of forced outage (see above).
The timing of these outages is, of course to a very large degree discretionary, and hence can be managed across a fleet to follow a demand profile. In principle, it should be possible to schedule outages in the summer. Figures 4.2 and 4.3 show the capacity utilisation of the nuclear and gas fleets. Through June to September, capacity utilisation averages around 75% and rarely goes above 80%. Therefore 20% of the fleet can be taken offline during those 4 months.
The end result is that an 85GW fleet largely made up of LWRs would, in fact, be operating close to its optimum, provided that the refuelling outages were aligned with the period of low demand.
Figure 3 – Monthly capacity from 2010 (coldest recent year) weather patterns with 2050 demand. Generating amounts are stacked and sum to demand.
Figure 4 – Daily capacity from 2015 weather pattern with 2050 demand.
It is quite feasible for the “tail” of capacity to be provided by oil fired power stations, without major impacts on emissions. Figure 5 illustrates a scenario, based on 2013 weather patterns, where capacity over 2,520GW / day (105GW average) is provided with oil fired power stations. Oil fired power stations have low capital cost, and an easy to store liquid fuel – they are therefore well suited to long standby periods and occasional use.
Figure 5 – Daily supply amounts assuming a mix of 85GW nuclear capacity, 20GW gas and the remainder diesel/bio-diesel
In this instance, the capacity amounts and factors are shown in the table:
Table 2 – Supply parameters for with diesel providing standby capacity. (The small difference in nuclear supply from table 1 is that this is based on 2013 weather only.)
Delivering 85GW of nuclear capacity
Initial LWR deployment
There is little point considering this model if physical and economic constraints make it infeasible to install the necessary capacity. This section will consider the viability of an 85GW nuclear fleet within the UK’s geography and grid characteristics.
The current plan (Wave 1) is to have approximately 20GW of nuclear capacity by 2030:
Table 3 – Wave 1 deployments
This deployment is possible, but the plan is perhaps optimistic. Given the issues regarding the EPR implementation at Hinkley Point, there is a chance that future EPR implementations – including Sizewell C – will need to be replaced by AP1000s, ABWRs, Hualong Ones or another design.
It should also be noted that Sizewell B will, even with a life extension to 60 years, be at, or close to retirement in 2050.
Auctions subsequent deployment
By the time it comes to commission Wave 2 – perhaps 2025 – there will be several accredited designs, probably including some Small Modular Reactors. The most cost effective way to deploy reactors will be for the Government to select site and obtain outline site approval, and then auction off the right to build and operate approved reactors. The Dutch Government has recently adopted this approach for offshore wind farms and has used this to cut bid prices significantly.
With this approach, the siting will be determined by the Government, but the actual builds to be deployed will be determined by the market. The build constraints will to some extent feed back into the site selection – as different technologies may be more suited to different sites.
Technologies for subsequent deployment
Debate about available nuclear technologies over a 30+ year timescale will inevitably be speculative to a degree, and will reflect the enthusiasms of the various proponents. In particular, the role of technologies such as liquid-metal cooled fast reactors (LMR) molten salt designs (in their various guises) (MSR) and Small modular designs (SMR) are open to debate.
Whilst the UK Government is attempting to accelerate the development of SMRs through development funding and competition, there are certain core constraints as to the ability of radical new technologies to play a major role, most specifically the timescales for licensing processes, and the need for any new technology to pass through a prototype/demonstrator phase before large scale commercial deployment.
This implies that a substantial roll-out of any new technology is unlikely until the mid-2030s (in line with the expectations of the international “Generation IV” consortium), and hence can play a major role only in the latter part of any build out at best.
Assuming a roughly linear build rate from today, approximately 3GW/year will have to be commissioned. To what extent this comprises existing LWR designs and to what extent it comprises of SMRs, will to some extent depend on the outcome of the Government’s development program and competition. It is prudent to assume that a “second wave” of LWR reactors will be required.
Considering each of the three LWR designs likely to be built (excluding EPR), there are reasonable extrapolations in terms of what can be expected in terms of capacity:
- AP1000 is capable of considerable “stretch” (SNERDI, in China is investigating options for 1700MW and 2100MW 3-loop derivatives), but Westinghouse ceded commercial rights for derivatives over 1300MW to China as part of the licensing deal there, resulting in the “CAP1400”. We’ll therefore assume that “second wave” AP1000s will be of 1300MW capacity.
- ABWR is already available in a 1600MW variant (proposed by Toshiba as opposed to Hitachi); we’ll therefore assume any ABWR build beyond those listed ahead are built at 1600MW capacity.
- Hualong-1 is a derivative of the 3-loop Framatome design of the 1990s, originally of 900MW capacity. There’s a practical limit on the degree to which such a design can be stretched; we will therefore opt for conservativism and assume 1300MW units.
- There is however political opposition to using Chinese based designs. Some of this is based on misplaced fears, but we would assume this might limit the Hualong to less than 10% of the overall market.
One other development option is worthy of note, albeit not making a material difference to the scale of deployment and available capacity – that is the “Reduced Moderation Boiling Water Reactor”. Currently under development by Hitachi, this is closely based on the ABWR design, with changes to the core design that leaves it operating on an epithermal neutron spectrum. This permits the “burning” of higher transuranics which cause issues in reprocessing of conventional LWR fuel. Replanning some or all of the assumed ABWR deployment with RMBWR would considerably ease waste management problems.
Location Options – a guideline scenario
We’re therefore left with a challenge to find a distribution of capacity that broadly fits with the UK’s geographical demand pattern and is both economic and has minimal environmental limitations. That implies:
- Generation close to demand, meaning a southern bias.
- Reuse of existing nuclear and generation sites where possible.
- Coastal or estuarine sites are to be preferred, where possible – this avoids the parasitic power losses and other constraints of cooling towers, or excessive heat dumping into rivers.
- Where existing inland sites are utilised (typically fossil fuelled), the new build should be no greater in thermal impact than the current stations.
- Where possible, there are operational gains in using similar technologies within a single site.
- As a minor consideration, within range of large centres of demand for district heating. This is not actually a severe constraint as the cost of the “local loop” will exceed the cost of main supply, even up to 100km distance.
Six of the current new build nuclear sites appear to have land area to accommodate six LWR units. These are specifically Bradwell, Hinkley Point, Sizewell, Hartlepool, Wylfa and Moorside. Our 2030 proposal assumed four LWR units/site; installing six LWRs at some or all of these sites would give us up to twelve of the eighteen additional units required.
Six units is a large station by world standards, but far from unknown – examples exist in Japan, Ukraine, Canada and France (at Gravelines, near Dunkirk). At these coastal sites, there is no lack of ocean cooling water – even if the tunnelling works to access the cooling water can be expensive (The cooling water for Hinkley C requires 9km of 7m-bore tunnelling).
Also in the 2030 proposal, we have further identified three additional sites apparently suitable for new build – Aberthaw, Killingholme and Tilbury/Isle of Grain. Expanding these sites to four LWR units would give us six further LWR units.
There are a limited number of former coastal coal-fired station sites that would seem to have potential – Lynemouth in the North East being perhaps the most obvious. Lynemouth would appear to have space for up to four LWRs.
Of a more contentious nature, two current nuclear sites could readily provide a home for 8-12 LWR units – Hunterston and Torness. This would, however required a significant shift on the political relationship between Holyrood and Westminster.
Subsequent LWR deployment
A Wave 2a deployment could consist of the following existing sites and technologies.
Table 4 – Wave 2a deployments
If EDF / Areva are able to sort out the issue of the EPR, and bring the cost down, then the Hinkley Point D build could be made up of EPRs. Assuming the Hualong-1 option, total nuclear capacity in Waves 1 and 2a come to 45.9 GW.
For the deployment if subsequent LWRs or novel SMRs, new sites are going to be needed. We assume a Wave 2b deployment of LWRs would only take place if SMRs do not make expected progress in the 2020s. A wave 2b deployment could consist of the following existing sites and technologies (new sites are in italics).
Table 5 – Wave 2b deployments
Making assumptions about capacity/site for the novel designs is inevitably more difficult, given that unit size ranges from the 50MWe of the NuScale unit to around 600MW for a twin PRISM unit, or the Moltex MSR.
A number of commentators have suggested that the SMRs can be deployed in small clusters near to cities to provide district heating as well as electricity. We think this is unlikely for two reasons:
- SMRs are likely to be grouped in clusters for security and operational reasons. Whilst a single 60MW NuScale unit is feasible, we would expect clusters to have at least 600MW of capacity.
- The main cost of district heating is the local loop. If a City builds a local loop, then it is possible to supply large quantities of hot water from some distance. For example, Bristol has recently announced plans for a district heating scheme – it would be quite feasible to supply it with hot water from Oldbury power station.
A typical small cluster of SMRs might consist of:
- A single twin PRISM unit (PRISM is design to deploy in pairs driving a single turbogenerator)
- 10-12 NuScale units
- 3 Westinghouse or Rolls-Royce SMR
- 5-6 mPower SMR units
- 2 Terrestrial Energy Molten Salt Reactors, if the technology progresses
There is indeed no reason why SMRs can’t be grouped in larger clusters – for example twenty of Terrestrial Energy’s IMSR600s would supply between 5 and 6 GW of electricity, whilst occupying less space than Hinkley C.
Sites for novel technologies.
Assuming SMRs make reasonable progress in development, they would be installed instead of the Wave 2b LWRs. We then need sites for around 40 GW of reactors, in clusters of various sizes. The first sites might include:
- Sellafield – where there is already a proposal to deploy two PRISM 300MW reactors as part of the UK’s waste disposal strategy.
- Trawsfynnydd – Site of a former MAGNOX station, there is considerable local support for this becoming a site for SMR development. Cooling capacity is inherently limited, however, being (a) located in a National Park, and hence unlikely to be allowed cooling towers, and (b) reliant on an artificial lake. A single 600MW cluster would be viable at the site.
- Dungeness – Dungeness was not included in the original list of sites for new reactors. The main reason given was cooling water constraints near a SSSI. There is however considerable local interest in maintaining a nuclear power station asset on the site. A higher efficiency SMR design – for a example a MSR or HTR (High Temperature Reactor) would alleviate the cooling issue.
The relative compactness of SMRs, even deployed as clusters, gives considerable flexibility in location terms. Two existing nuclear sites have potential in this context – Heysham and Dungeness (see above). Heysham is an awkwardly shaped site, but is not constrained in cooling water capacity.
Our best prospects for siting further clusters are current and recently vacated coal-fired sites (albeit Didcot is lost to housing development). Typical former coal-sites have supported stations in the range 1000-2000MW, hence a reasonable assumption of 2GW sites is not infeasible, utilising sites such as Rugely, West Burton, Eggborough, etc. Unfortunately, these sites tend to be associated with the former coalfields, and are hence further north than is ideal. Hence potential early sites for SMRs could include:
Table 6 – Wave 2b SMR deployments
In practice, a mixture of “wave 2b” and SMR deployment is perhaps the more probable option, allowing for a gradual cutover as the SMR market becomes established. That results in a total supply position of somewhere between 66,000 and 86,000 MW.
Even with the above list, we could be up to 20,000MW short for our 2050 target. Finding sites may be difficult – as it has been for onshore wind farms. Perhaps nuclear could follow the wind farms offshore. At least one developer of SMRs has proposed siting clusters of reactors offshore, on artificial islands. The island would be situated about 12-20km offshore, in water of 5-10m depth (at high tide).
Initially, the island wall would be built, 20m above the sea bed, to provide a sheltered marina. Then, complete SMR units – of about 250MW electrical capacity – would be floated in at high tide. The units would be on barges, complete with nuclear island, steam turbines, and all associated support facilities. Once the barges are in place they are ballasted down, and the remaining marina space is filled in with sand – apart from the harbour. Any reactor (up to a height of about 18m) or fuel store is below the island level, protected from the sea by at least 30m of rock and sand, as well as its own protection of metal and concrete.
To accommodate Molten Salt Reactors, the island would be about 230m wide (excluding the sloping sea wall), with a length of 160m plus 160m per Gigawatt (of electrical output). Hence a 4GW island would be approximately 800m long, an 8GW island 1,440m. A Pressurised Water Reactor island would be quite a bit larger. The cost of the “island-works” in 5-10m of water depth (at high tide) would range from approximately £200 million for a 1GW island to £500 million for a 8GW island.
Siting the reactors in offshore islands has numerous advantages:
- The cost of building the island is cheaper than cost of onshore “nuclear land”;
- Access to cooling water is easy – the whole island cost may actually be cheaper than the tunnelling works for Hinkley C;
- The entire power unit – including the nuclear and generating sections – can be factory built and floated to the site;
- For most designs, the actual reactor core is partly or totally below the high water wave level. This means that unlimited passive cooling can be provided to provide true “walk away safety”;
- There are no neighbours to upset – and the sites comply with existing Government guidelines on nuclear power station siting;
- There are a number of suitable sites on the East England coast, from the Thames Estuary up to Lincolnshire. (On the west coast, the heavy tidal ranges would increase the cost of the island – the sea wall cost rises with the square of the wall height).
- The sites are near enough to shore that HVDC connectors are not required – though they could be built to access more distant connection points on the East Coast or on the continent;
- If the islands need to be dismantled and the reactors decommissioned, then the marina sand can be pumped out, and the barges can be towed to the decommissioning factory.
- An island can be easily expanded. If it starts with 1GW of capacity, it can be lengthened to 8GW subject to appropriate sea bed depths;
- This approach also enables the export of “SMR islands” across Europe and beyond. 250-500MWe barge units could be manufactured in a former ship works and deployed on the coast near to demand centres.
It would however require extensive validation by the regulators in terms of resistance to extreme weather and tidal situations, as well as security. Each site could house 4 to 8GW of reactors, easily bringing the 85GW target within reach.
Figure 6 – Simplified cross sectional view of a nuclear island – “generic” design for SMRs. (One service barge will support a few GW of reactor capacity)