Essentially EDF and Areva thought that since they were French and operated lots of nuclear power stations, they knew how to build them. This was wrong. You’re probably familiar with First of a Kind (FoaK) but there is also such a thing as First in a While. They made a number of mistakes early on in the EPR programme. (For reference, there have been six plants EPRs where construction has started. In order: Olkiluoto in Finland, Flamanville in France, Taishan 5 and 6 in China, HPC 1 and 2 in the UK) Sizewell would be numbers 7 and 8.
A brief aside on over-budget and over-time here. These are the same thing on complicated major infrastructure projects. Virtually all the costs are labour costs and these are all specialised skills so when they’re not working, you don’t just tell the contractors to go work on something else, you need to pay them the whole time they’re mobilised. A schedule delay = being over budget.
Most significant was the start of Olkiluoto construction before the detailed design was done. Starting construction before detailed design is very common in construction and it’s really more of a balancing act than a binary decision but in this case it was a very bad idea. There were substantial periods of time where thousands of people were being well paid to do nothing while Areva and other partners were working out design issues that held up the build.
Second was some absolutely unforgivable fuckery going on at the main forging works in France. The biggest single item in a PWR reactor is the forged pressure vessel which is made in enormous pieces, these pieces are then painstakingly welded together – each weld may take 100 passes in between each the weld is examined, cleaned, and the work pieces are heated up. There were documents being forged about quality control, really really bad stuff. As a result, the Flamanville vessel lid needs to be replaced very early in its life since it has defects that mean it will not last the 80 year design lifetime.
Third, there are non-up-to-specification welds in the main cooling lines at Flamanville, they’re in a place that is no longer reachable and so EDF has had to develop custom welding robots to fix them.
The two Chinese plants were somewhat late but not so over-budget. Part of the reason for that is China has a big nuclear construction programme so as soon as they needed to halt while a design issue was worked out, they sent all their nuclear qualified on-site staff to work on other projects.
Third: why is the agreed strike price for HPC so high?
First, the capital cost is high. The estimated capex was based on the cost it actually took to build Flamanville and Oikiluoto at least the latter of which has actually been built properly and is now entering final commissioning. That does mean that the capex estimate is more accurate than the much lower costs estimated before construction started on those plants.
Second, the strike price accounts for the entire construction risk sitting with EDF.
Third, EDF has had to finance the whole cost itself. If you built the same project using government borrowing costs the project would be much cheaper. (However this is an economically problematic view since government borrowing costs being low relies on a risk transfer to citizens, a matter on which I disagree with the National Audit Office).
Fourth: How’s HPC construction going?
So far, very well. The second unit is being built faster and more cheaply (remember – same thing) than the first. The timing is such that if you want to get maximum nth of a kind savings from building Sizewell C, you need to make a decision soon so that you can organise moving staff from one to the other. It is worth noting that econometric studies of nuke construction costs have shown that series building saves a huge amount of money so it would actually make no sense for the UK to build any non-EPR designs unless they have radically different characteristics like SMRs to justify it.
Finally: is it therefore the right decision to build Sizewell C?
Honestly. I do not 100% know and I am deeply suspicious of anyone who does. They either know more about it / have thought about it more than I have, and of course I accept that there are people who do and have. Or… they haven’t but are nonetheless sure of their position. My leaning is that long term government thinking should be aimed at actions that are near-optimal / low-regret in the greatest number of cases. Given the uncertainties around CCS as an alternative and the actual climate impact of methane leakage on the way to the CCS plant, the absolute over-riding-everything need to deal with our emissions and to do so as fast as we can, and the long construction time scales, I think it probably is the right decision to build at least this plant.
A decision on a third two-unit EPR does not need to be made until later and at that point we will know more about future technology developments. It may be that at that point we realise that not only do we not need a third but that we didn’t really need Sizewell and HPC either and therefore we will have spent more by 2050 to decarbonise than was optimal. Totally possible! The objective though is not to optimise in a fragile way but to create as many somewhat-optimal pathways as possible.
(disclosures: I make part of my income from working on wind, solar, battery and lately hydrogen projects. I've never worked on a nuclear project.)
Essentially EDF and Areva thought that since they were French and operated lots of nuclear power stations, they knew how to build them. This was wrong. You’re probably familiar with First of a Kind (FoaK) but there is also such a thing as First in a While. They made a number of mistakes early on in the EPR programme. (For reference, there have been six plants EPRs where construction has started. In order: Olkiluoto in Finland, Flamanville in France, Taishan 5 and 6 in China, HPC 1 and 2 in the UK) Sizewell would be numbers 7 and 8.
A brief aside on over-budget and over-time here. These are the same thing on complicated major infrastructure projects. Virtually all the costs are labour costs and these are all specialised skills so when they’re not working, you don’t just tell the contractors to go work on something else, you need to pay them the whole time they’re mobilised. A schedule delay = being over budget.
Most significant was the start of Olkiluoto construction before the detailed design was done. Starting construction before detailed design is very common in construction and it’s really more of a balancing act than a binary decision but in this case it was a very bad idea. There were substantial periods of time where thousands of people were being well paid to do nothing while Areva and other partners were working out design issues that held up the build.
Second was some absolutely unforgivable fuckery going on at the main forging works in France. The biggest single item in a PWR reactor is the forged pressure vessel which is made in enormous pieces, these pieces are then painstakingly welded together – each weld may take 100 passes in between each the weld is examined, cleaned, and the work pieces are heated up. There were documents being forged about quality control, really really bad stuff. As a result, the Flamanville vessel lid needs to be replaced very early in its life since it has defects that mean it will not last the 80 year design lifetime.
Third, there are non-up-to-specification welds in the main cooling lines at Flamanville, they’re in a place that is no longer reachable and so EDF has had to develop custom welding robots to fix them.
The two Chinese plants were somewhat late but not so over-budget. Part of the reason for that is China has a big nuclear construction programme so as soon as they needed to halt while a design issue was worked out, they sent all their nuclear qualified on-site staff to work on other projects.
Third: why is the agreed strike price for HPC so high?
First, the capital cost is high. The estimated capex was based on the cost it actually took to build Flamanville and Oikiluoto at least the latter of which has actually been built properly and is now entering final commissioning. That does mean that the capex estimate is more accurate than the much lower costs estimated before construction started on those plants.
Second, the strike price accounts for the entire construction risk sitting with EDF.
Third, EDF has had to finance the whole cost itself. If you built the same project using government borrowing costs the project would be much cheaper. (However this is an economically problematic view since government borrowing costs being low relies on a risk transfer to citizens, a matter on which I disagree with the National Audit Office).
Fourth: How’s HPC construction going?
So far, very well. The second unit is being built faster and more cheaply (remember – same thing) than the first. The timing is such that if you want to get maximum nth of a kind savings from building Sizewell C, you need to make a decision soon so that you can organise moving staff from one to the other. It is worth noting that econometric studies of nuke construction costs have shown that series building saves a huge amount of money so it would actually make no sense for the UK to build any non-EPR designs unless they have radically different characteristics like SMRs to justify it.
Finally: is it therefore the right decision to build Sizewell C?
Honestly. I do not 100% know and I am deeply suspicious of anyone who does. They either know more about it / have thought about it more than I have, and of course I accept that there are people who do and have. Or… they haven’t but are nonetheless sure of their position. My leaning is that long term government thinking should be aimed at actions that are near-optimal / low-regret in the greatest number of cases. Given the uncertainties around CCS as an alternative and the actual climate impact of methane leakage on the way to the CCS plant, the absolute over-riding-everything need to deal with our emissions and to do so as fast as we can, and the long construction time scales, I think it probably is the right decision to build at least this plant.
A decision on a third two-unit EPR does not need to be made until later and at that point we will know more about future technology developments. It may be that at that point we realise that not only do we not need a third but that we didn’t really need Sizewell and HPC either and therefore we will have spent more by 2050 to decarbonise than was optimal. Totally possible! The objective though is not to optimise in a fragile way but to create as many somewhat-optimal pathways as possible.
(disclosures: I make part of my income from working on wind, solar, battery and lately hydrogen projects. I've never worked on a nuclear project.)