Apologies for the delay on coming back to this. I ran the WEIS no. 15 example case (without optimization) and find that RAFT and WISDEM are reasonably well aligned with mass estimates:
floatingse.platform_mass,output,kg,16880841.200742528,
raft.properties_substructure mass,output,kg,16977818.323086586,Substructure mass
That is a 0.6% difference.
I didn’t run your optimization, but I suspect there that the issue may be similar to your other LCOE question. I think your optimization has not fully converged and may be at an intermediate stage where constraints are violated, which WISDEM and RAFT may accommodate differently. Without seeing more details of the output, it is difficult for me to comment further.