For tokamaks, there's an idea that we can take all the relevant size and plasma parameters for a bunch of different experiments, run a fit on them, and come up with a set of scaling factors that will tell us how well a reactor of size x is going to perform. The prediction is that the bigger the reactor, the better things get, which is why ITER is the biggest tokamak yet.

People call it 'wind tunnel scaling', so I presume this is an empirical method borrowed from aerodynamic modeling. Nowadays we can actually do computational fluid dynamics to some extent, but a full-on electromagnetic plasma simulation remains intractable on the macroscopic/machine scale. And we remain limited on theoretical approaches, so we are stuck with actually having to build the darn things before we know they'll work.

I'm definitely not an expert, but building new fission plants to modern standards seems to be substantially more difficult than previously thought.

The only new fission plant being built in Western Europe, Olkiluoto 3 in Finland, was supposed to go online in 2010 but is still unfinished: http://en.wikipedia.org/wiki/Olkiluoto_Nuclear_Power_Plant#U...

Estimates suggest it may not be done until 2020, and the budget has overrun to about $10.6 billion USD versus the original cost estimate of about $3.75B USD. (This reactor is now more expensive than the Large Hadron Collider!)

When a fission plant costs $10B+ today, I wonder how much the first working fusion plant will cost -- if we ever get that far.

I did some scientific computing for OL3 in 2009-10 and the issue you're referring to is a different one. Back in the 80's, there was around 5 new nuclear plants built in France, each year. It powers 75% of our electric grid and most of these plants are still going to be here till 2020-2040. Now most people who conceived and designed those plants are retired or almost retired. The new EPR plants (OL3, FA3) suffered consequently, the project management and the engineering is pretty much fucked up (personal point of view, I didn't had much insight). Also Areva consciously sold the plant for a low price in order to win the auction, they knew from day 1 that it would be delayed but probably not that much.

Now I was referring to the inherent difficulty to scale fusion up due to the underlying physics. So when the article says:

> The next steps for the dynomak are straightforward. The experimental device [...] is about one-tenth the size that a commercial dynomak fusion reactor would be. [...] the group hopes to construct HIT-SIX [...] that will be twice as large as HIT-SI3. >At that size, things start to get interesting, says Sutherland. If imposed-dynamo current drive works well in HIT-SIX, he’ll be “much more confident going forward that our development path will be successful,” he says.

I don't know if it will be that straightforward because they're building a 1/10 prototype to build a 1/5 proto and then if all goes well they could be more confident about building a full scale plant. Does someone know how scaling up 1/10 → 1/5 → 1/1 is or isn't such a big issue ?

You forgot Flamanville 3 in France: http://en.wikipedia.org/wiki/EPR_(nuclear_reactor)#Flamanvil.... Same problems though (huge delay, budget increase, ...)

And Hinkley C in Somerset, UK http://www.theguardian.com/business/2014/nov/20/hinkley-poin... (again, delays etc - currently only earthworks are going on)

Also not an expert. My impression was that the high cost of fission plants can be attributed to safety features that aren't required with fusion. (And arguably aren't required -- to the extent society demands them -- with fission plants either.)