Estimates of Chernobyl's costs have been exacerbated by a number of factors, not the least of which was a number of ridiculous things that were done to cover up the mess.
The $150 billion Katrina estimate is ignoring a number of factors including things like... loss of life, which you know, tends to impact the costs of things.
Just the Federal government has spent >$100 billion and nobody in the area even tries to pretend that they've restored it to its previous state or undone the economic harm. Insurance companies, as you said, have paid $41 billion, so there you're at $141 billion and you've barely scratched the surface as compared to the things factored in to that $235 and $257 billion figures.
But let's cast all that aside. Let's assume the cost is 2-3x Katrina. The incident rate is obviously not that high (there's a reason basically all nuclear plant failures trace back to a design from the 60's), but I'm sure you can plug that in.
There is a pretty handy Wikipedia page with details for Nuclear Plant disasters: http://en.wikipedia.org/wiki/Nuclear_and_radiation_accidents...
Now, I haven't done the math for how many aggregate plant-years of operation we've had, but let's say your average plant has been in operation for 10 years (which is definitely low-balling it in the US, but the mere fact we run these old nuclear power plants is exactly why they are so much more likely to have problems). That's 4360 plant-years. The Wikipedia page has 20+ incidents listed, but that includes a lot of cases that are tiny compared to the disasters you are thinking in terms of (and of course, ironically, anyone who is insuring these things will tell you that those smaller incidents are the primary drivers of the costs in an insurance policy). I count 5 cases throughout history with costs >$1 billion. That works out to a very exaggerated rate of one incident every 972 plant-years.
So, with all those "tie your hand behind your back" factors, a premium of $50 million/year would cover costs very well and leave the insurer rolling profit margins that'll probably create a congressional investigation. That's a lot of money, but compared to the economic output of a power plant, isn't prohibitive.
Realistically though, newer reactor designs, particularly post-3 mile island, have a much lower incident rate than their predecessors and much better mechanisms for containing damage. If insurers were driving the design of plants, it'd probably be even lower (or costs would drop). Most insurers would probably also adjust their premiums based on the age of plants, which at some point would make it more cost effective to build a new one than to keep operating an old one. They'd also reduce costs by establishing rules that would limit their liability in circumstances where the failure was due to some other party not managing their responsibilities.
I'm curious what the real numbers are, but you can see that even being grossly unfair, insurance costs for disasters are not going to make nuclear plants economically non-viable.