I am running a transient simulation. After exporting the solution field to CSV, I compare U:0, U:1, U:2 through time to see if they changes, and they do! I think they should not. In the screenshot below, a and b are from consecutive time frames
small changes in the velocity field are very usual.
#1 Transient effects / turbulence. Flow is never perfect laminar and continuous. Therefore we expect to have small variations due to turbulence, #2 Numerics. Since all equations are being solved numerically it’s also expected to have some changes.
Therefore I would always suggest averaging your last n results of a transient simulation to get a good average results