Odd things happen, such as Chernobyl and Fukushima, yes.
Well, let's be honest: neither of those incidents were odd or unforeseeable. In the case of Chernobyl, we had an experimental reactor (and I don't mean it was new - I mean it was specifically built for them to screw around with and see what happens) designed with a highly positive void coefficient. It was an insane design that was not passively safe and it was purposely operated in a reckless manner. The "accident" that took that place down happened when they shut off the already limited safety features and ran more experiments. Keep doing that over and over in a design that isn't passively safe and you almost can't help but have it end in disaster. If nothing else, any rational person could easily see that what they were doing was dangerous as Hell. And the Soviets knew what they were doing was dangerous as Hell which is why they did it there and not next to, say, Moscow for instance.
In the case of Fukushima, the plant design's manufacturer (GE) identified design flaws in the plant's containment measures back in the 1970s. And they came up with a remediation plan and published it to everyone running that design. In the 1970s. And the company operating the plant at Fukushima chose not to do what GE told them they needed to do in order to ensure containment in the event of a catastrophic failure. And the regulators in charge of ensuring the plant was operated safely allowed them to do that. So the plant ran for decades with a known design problem and nobody did anything about it. So again, this wasn't exactly a surprise that as soon as something went wrong, bad stuff happened.
Ain't no magic here: if you run known-unsafe designs, you're risking bad things happening. If you run safe designs, then catastrophic failures do not (and, physically, cannot) result in catastrophic consequences.