I'd expect these shutdowns to be more-or-less randomly distributed across calendar years[1], and in aggregate, accounted for, on a macro level, in supply/demand projections.
[1] The examples you cited suffer from confirmation bias - a plant shutting down due to a fire when there isn't already a chip shortage, would not really make the news. And if it did, you wouldn't be actively looking for it!
