This person studied many thousands of large scale engineering projects and the pattern of these projects overrunning time and costs budgets is universal according to him. It's not a local problem specific to the US either. It happens everywhere.
A few of his observations were that renewable projects are among the best performing large scale energy projects currently and nuclear projects among the worst. The big differences between the two are:
- on site vs factory based manufacturing.
- complexity
Nuclear plants are built from scratch in the field. This has lots of risks, requires lots of ad hoc problem solving, and any emerging problems lead to costly delays. That also means that a lot of learning effects are lost because there are a lot of contractors involved that have to solve a lot of site specific issues and in between projects years/decades can pass so the next project has to re-learn a lot of things. And as each site is slightly different there are also site specific risks and uncertainty. The learning effects are negative because each project inherits the mitigations from the previous projects and then has to add completely new solutions based on any new problems.
With renewables, most of the difficult stuff happens in a factory. That factory produces lots of cheap modular units that are than shipped and assembled on site. Each unit is relatively cheap. Assembly is relatively uncomplicated. With a wind tower for example, the components are large (tower, turbine, blades, foundation) but mostly pre-assembled. So, on site assembly is relatively straightforward and does not require a lot of problem solving. With solar it's even better because the parts are more plentiful, smaller, and assembly does not require a lot of heavy equipment or site preparation. Doing this at scale means there is a large positive learning effect.
So, estimating a wind or solar project is possible with a high degree of accuracy. These projects can be scaled up or down as needed. With Nuclear it so far is the opposite.
As for modular nuclear reactors, which the article barely talks about, this is a promising but as of yet unproven way to produce reactors in factories. It's promising because it duplicates some of the advantages of renewables. It's unproven because we won't know it works until lots of these have been assembled and deployed. Which is a thing that seems to be blocked on a lot of mundane things like regulations, approvals, investor enthusiasm, etc. And there's also the notion that the nature of nuclear reactors is that so far bigger seems better in terms of efficiency and overall cost. Which goes against producing lots of small reactors that are two orders of magnitudes smaller than their traditional multi GW counterparts. Likely at least for the first few ones, cost and time budget estimates are probably not that accurate.
Michael Liebreich recently interviewed a researcher that made some interesting points about the phenomenon of big public projects overrunning cost: https://www.cleaningup.live/ep128-prof-bent-flyvbjerg-how-bi...
This person studied many thousands of large scale engineering projects and the pattern of these projects overrunning time and costs budgets is universal according to him. It's not a local problem specific to the US either. It happens everywhere.
A few of his observations were that renewable projects are among the best performing large scale energy projects currently and nuclear projects among the worst. The big differences between the two are:
- on site vs factory based manufacturing.
- complexity
Nuclear plants are built from scratch in the field. This has lots of risks, requires lots of ad hoc problem solving, and any emerging problems lead to costly delays. That also means that a lot of learning effects are lost because there are a lot of contractors involved that have to solve a lot of site specific issues and in between projects years/decades can pass so the next project has to re-learn a lot of things. And as each site is slightly different there are also site specific risks and uncertainty. The learning effects are negative because each project inherits the mitigations from the previous projects and then has to add completely new solutions based on any new problems.
With renewables, most of the difficult stuff happens in a factory. That factory produces lots of cheap modular units that are than shipped and assembled on site. Each unit is relatively cheap. Assembly is relatively uncomplicated. With a wind tower for example, the components are large (tower, turbine, blades, foundation) but mostly pre-assembled. So, on site assembly is relatively straightforward and does not require a lot of problem solving. With solar it's even better because the parts are more plentiful, smaller, and assembly does not require a lot of heavy equipment or site preparation. Doing this at scale means there is a large positive learning effect.
So, estimating a wind or solar project is possible with a high degree of accuracy. These projects can be scaled up or down as needed. With Nuclear it so far is the opposite.
As for modular nuclear reactors, which the article barely talks about, this is a promising but as of yet unproven way to produce reactors in factories. It's promising because it duplicates some of the advantages of renewables. It's unproven because we won't know it works until lots of these have been assembled and deployed. Which is a thing that seems to be blocked on a lot of mundane things like regulations, approvals, investor enthusiasm, etc. And there's also the notion that the nature of nuclear reactors is that so far bigger seems better in terms of efficiency and overall cost. Which goes against producing lots of small reactors that are two orders of magnitudes smaller than their traditional multi GW counterparts. Likely at least for the first few ones, cost and time budget estimates are probably not that accurate.