On the blow-up mechanism of moving boundary problems