This is one of the weird occasions where EnergyPlus reports a "severe" error, but it still allows the simulation to proceed. Note that you do obtain results in BEopt. In most cases, a "severe" error causes EnergyPlus not to proceed with the simulation in order to avoid producing erroneous results. In this case, the error seems reasonable to ignore -- it should only result in a slightly different initial temperature for the unconditioned basement, which is unlikely to lead to much of a difference in simulation results compared to if EnergyPlus had converged successfully.

It's possible that BEopt 2.8 was suppressing this EnergyPlus error and BEopt 3 is not anymore; we'll have to take a deeper look.