slip connection at top of non-load bearing cmu partition

huangyhg

slip connection at top of non-load bearing cmu partition
i am designing a large pre-engineered steel building with frames that span about 133 feet (ice hockey rink).  part of the building has a mezzanine, really a second floor, with a sports bar and an exercise room.  the non-load bearing cmu walls aroung these spaces extend to the roof above (the architect shows them butting to the roof, which of course should not be done).
i want to devise some sort of slip connection to brace the top of these walls up at the roof, but i don't know how much the roof will deflect (pre-engineered, but not designed yet).  if i assume the roof could deflect as much as span/180, this would be almost 9 inches!!!
has anyone else seen this before?
daveatkins
i would suggest designing the wall as a true cantilever (no fixity at roof) or maybe you could embed a steel plate in the top of the wall and use a slotted connection at the roof.
are there any perpendicular walls that you could use as bracing?  depending on the horizontal span of the masonry wall that you are concerned with, you may be able to design it as braced (pinned) on three sides (bottom, and two sides where walls frame into it).  
you could probably devise a slip connection using a bent plate angle with slotted holes and teflon washers, provided you have some good steel at the roof to which to connect.
i agree with part of motorcity's response, design your structure standalone, but then stay away from the metal building entirely.  the vertical deflection may be as much as 9 inches, but what about the lateral deflection?  do you happen to have a copy of the metal building spec that would list their sidesway design criteria?  unless the exterior walls are cmu, i seriously doubt the metal building engineers kept cmu cracking in mind when they designed for sidesway!
i have ran into this before.  call out a max. deflection that you can live with for the pemb rafter.  then attach a bent plate to the underside of the rafter to accept the top of the cmu wall; like a deflection track in a metal stud wall.  the "deflection track' should come in 2 parts -a bent plate attached to the rafter before the cmu wall is ran all the way up, and flat plates welded vertically to the bent plate crating the other leg of the track.  you can provide some long slots with expansion bolts in the leg of the tracks, but i feel that only provides a place to bind and i usually don't.  provide a detail for the pemb supplier so that they can properly brace the bottom flange of the rafter.
i basically did what ntpe suggested (the drawings went out yesterday).  i left a 10-1/2" gap above the cmu walls (i allowed for purlin deflection as well) with a bent plate each side of the wall welded to each purlin (will probably be 5' oc) or purlins between purlins when the wall is parallel to the purlins.
but pmr06 raises a good point as well, and i'm not sure how to deal with that.  i do think that vertical deflection of the frame is more likely than lateral deflection of the entire building, however.
daveatkins
re  
good point, ntpe.
i plan to look at the pemb shop drawings when they arrive--if the deflection is < 10-1/2", i can revise the detail.
daveatkins
ntpe - we do what you have suggested.  the only tough thing is that the pemb designers really get testy when you tell them how much drift, how much deflection, etc. to use.
jae - you are rignt about the pemb designers, sometimes they don't want any restrictions, but they are not the eor. i have rejected pemb shops for noncompliance with the contract docs, this usually brings tham back into line.