Uh...perhaps you're unaware all three reactors shut themselves down in response to the quake, just as they were designed to do. The problem wasn't that the reactors didn't shut down; the problem was even a shut down reactor generates significant residual heat for almost a month after it's shut down. This heat may be only 1%-3% of total reactor power, but if not removed it's more than enough to melt the reactor core over time.
Removing this heat is the job of the Residual Heat Removal System, a fancy name for something that essentially just circulates coolant through a shut down reactor. The coolant picks up heat in the reactor, goes to a heat exchanger to get rid of that heat, and repeats the cycle. But these pumps require power, and power was knocked out (a) when the plant shut down, (b) when local utility power was cut due to the quake, (c) when the tsunami destroyed the backup diesel generators, and (d) when the emergency batteries finally ran out of juice.
So, your initial premise that Fukushima was not shut down correctly is...well, incorrect. It was shut down correctly, automatically. The meltdowns occurred not due to human error but due to a disaster that ran the gamut of all the safety systems and backups and contingencies that the plant was designed to handle. When you inflict a situation on an engineered system that exceeds its design specifications, failure is not only likely, it is the expected outcome. Build a higher seawall? There's still a tsunami that could top it, given the proper -- if unlikely -- circumstances. Build it for a 9.0 quake? You could still get a 9.1 quake. And even if you mitigated all of these events, what about asteroid strikes? Space aliens? No matter what you do, it is impossible to eliminate all risk from any complex system. You design for the most probably cases, add in a healthy safety margin, and that's the best you can do.