I agree with Rich and advise you to continue the model past the bottom of your tower, to a logical point of isolation or a boundry condition which can clearly be defined. A so-called anchor (3)fixed linear vectors and incidental rotational stops would fit the bill.

Also the period of the tower and hence the downcomer strapped to it will probably be different than the structure which the "anchor" is attached to hence consider SAM effects (Seismic Anchor Motions)

------------------
Best Regards,

John C. Luf