%tim_act certainly is a really bad thing and indicates a problem.

If you have 88 DB volumes, I think that means TSM can in theory start up to
88 I/Os to the DB, which could certainly cause excessive head thrashing.
The problem with the EMC box is that it's hard to tell how those I'O's map
to physical disks.  You may need to get EMC on site to figure out how to
separate that I/O out in the box, or to determine if additional cache in the
box will help..

Moving to 8 files of 10 GB in size will NOT do any compaction or reorg, but
it has the advantage of reducing the number of I/Os that may occur
concurrently, which in your case is probably a good thing.  If you could
(with EMC's help) make sure those 8 files are on physically separate disks,
that should help, unless the problem is that some other host is causing the
I/O thrashing in the box.

ANother thing to check:  q db f=d
What is your DB cache hit ratio?  You can have a case where the OS shows
memory is NOT overcommitted, but still have your TSM buffer pool too
small...