Dear Brendan,
I believe the equation in HydroDyn is correct because this is the inertia about the node at the end of the cylinder, not the inertia about the center of mass of the cylinder.
Before the improvements to HydroDyn in v2.6, the inertia of the filled fluid in HydroDyn assumed the filled fluid was a thin rod, neglecting the rotational inertia within the cross section. This was discussed in the following forum topic and mentioned in the draft HydroDyn User’s Guide and Theory Manual: AddCLin for OC4 Semisubmersible - #3 by Jason.Jonkman.
If both v2.5 simulations match the non-fluid-filled v2.6 solution closely, then I would guess what happened is that the steel-only inertia in the model with filled fluid was corrected to account for the missing inertia term in HydroDyn. Now that the fluid fill inertia has been improved in HydroDyn within v2.6, the steel-only inertia should be adjusted accordingly to ensure that the overall platform inertia of the steel + ballast is correct. It looks like we should update the NRELOffshrBsline5MW_OC4DeepCwindSemi_ElastoDyn.dat file in the OpenFAST r-test in v2.6 and newer accordingly, once the correct steel-only inertia has been calculated.
Best regards,