It really depends on the type of test. If, for example, you’re verifying a certain software requirement is met, running a test 3x won’t give you any information. If you’re running a process that could vary based on different factors, then multiple executions are warranted.
I don’t know of any regulation or guidance (maybe GAMP has something?) but 3x does seem to be the de facto standard when multiple executions are performed.
Generally what I’ve seen (and there are plenty of valid reasons where this could deviate) is IQ and OQ are single pass and PQ is done over 3 passes (each for whatever configuration is done).