Are you using the energy method for the low flow bridge method?
Another issue, when you add a bridge RAS creates two additional cross sections inside of the bridge. So even if the water doesn’t touch the bridge there are two more cross sections. This could affect your results.
Ok. I just noticed you said unsteady. Unsteady creates bridge curve tables that are used during the solution. The answer inside the bridge is computed later by the post processor. You can get slight differences between the two.
Look to make sure the bridge curve looks smooth. Upping the number of curves or the number of points per curve might reduce the discrepancy.