Use observed stops to size the problem
Record where stops originate, how long they last and whether upstream or downstream machinery can continue safely. A buffer that covers a frequent short label-roll change may be useful; a buffer sized around an unresolved ten-minute process fault may consume space without improving sustained output.
