You may be getting confused regarding Gage Repeatability/Reproducibility and Reliability measures. The only instances where GR&R would be done on a time based situation would be if an instrument or device was actually measuring time (e.g. stop watch, intermal machine counter/clock, etc.). I have never actually witnessed a GRR performed on such devices; and don't know what "standard" would be used.
Machine downtime is a reliability metric. The statistic there is, for example, MTBF (mean time between failures). Usually, the precision of this metric, whether in minutes or hours or days, is not a concern.
We are tracking total minutes per changeover (from end of one job to beginning of packing of another) on a continuous mfg machine. This measure has been traditionally calculated by (24 - Total lbs produced / avg lbs hr ). For example: Total lbs produced = 47000. Avg lbs hr = 2000. So 24-23.5 = .5 or 30 min. That is our downtime for the changeover.
First, let me correct my previous post: MTTR (Mean Time To Repair) is a metric for measuring downtime, not MTBF. SORRY!
Second, billybilly, is that if that is your method for measuring downtime, and it is consistently so, then that is fine - there is no GRR issue here. When keeping track of MTTR, you will create a distribution, e.g. weibull or lognormal, and then determine you confidence bands. To be accurate you should use Reliability software. But you don't have to. Simply create a trendline and note when changes are made to reduce downtime (or its variability).
Sorry I don't have the time right now to better explain this. Maybe I'll bet a chance later to create a new post on this subject.