DPOS vs OPUS Discrepancy

I have been working on a large route project for a while now here in Ohio. We did static observations on all of our primary control points. The initial plan was to do 2 separate 2.5 hour sessions on each point. Using OPUS to process, on a lot of our points we didn't meet our quality criteria (<0.10' vertical) for the two sessions on the same point. We then switched to 4 hour sessions and that problem went away. Vertical is the weakest link with GNSS, long observations are needed.