Nuclear plants take years to build and have shitty failure scenarios. Best to stopgap coal until renewables take over with combined cycle natural gas plants, fueled by cheap natgas from fracking.
It's not perfect, but those natural gas plants will still be kept around as peaking plants and they're an order of magnitude cleaner than coal plants (as well as producing much less CO2 per unit of power generated).
> Nuclear plants take years to build and have shitty failure scenarios
That's only because we won't build any newer designs that don't have shitty failure modes and insist that we keep old reactors with very shitty failure modes online, though.
I don't disagree that we can build better designs. I disagree that we can build them quickly and cheap.
According to the OECD:
"As nuclear power plants are complex construction projects, their construction periods are longer than other large power plants. It is typically expected to take 5 to 7 years to build a large nuclear unit (not including the time required for planning and licensing)."
"Between 2002 and 2008, for example, cost estimates for new nuclear plant construction rose from between $2 billion and $4 billion per unit to $9 billion per unit, according to a 2009 UCS report, while experience with new construction in Europe has seen costs continue to soar."
All the modern design proposals are to mass-produce smaller units -- sometimes, completely sealed units -- which would then be installed several to a site, and maybe sent back to the manufacturer to be refurbished at periodic intervals. This is the "save money by mass production" economies-of-scale approach, as opposed to "save money by being more efficient with large custom-built plants" economies-of-scale approach which has failed the industry and led to many of those ballooning costs.
(Whether we have the regulatory wherewithal to go with that plan is another matter, but the very idea of reusing the same design over and over does help keep compliance costs down, and you can focus on site-selection costs.)
How does that compare with other sources? 5 Years and a few billion doesn't seem that bad to me. Even plans for building a few thousand new homes can be over the next ten to fifteen years.
Surely the main cause of failure is human rather than the actual design. Nuclear stations may be safe, but that comes at an increasing cost in money and complexity. How do we deal with the institutional failings that allowed us to build "old reactors with very shitty failure modes". Stop blaming it on the tech. I want reactors that are simple enough to be cheap and easy.
Humans will always make mistakes, even within the design process. Nuclear is a dangerous technology to make a mistake with.
Diablo Canyon Power Plant was partially built backwards, by mistake. It's absurd.
The company updated its plans and added structural supports designed to reinforce stability in case of earthquake. In September 1981, PG&E discovered that a single set of blueprints was used for these structural supports; workers were supposed to have reversed the plans when switching to the second reactor, but did not. According to Charles Perrow, the result of the error was that "many parts were needlessly reinforced, while others, which should have been strengthened, were left untouched." Nonetheless, on March 19, 1982 the Nuclear Regulatory Commission decided not to review its 1978 decision approving the plant's safety, despite these and other design errors.
It's not perfect, but those natural gas plants will still be kept around as peaking plants and they're an order of magnitude cleaner than coal plants (as well as producing much less CO2 per unit of power generated).