Need a book? Engineering books recommendations...

Return to index: [Subject] [Thread] [Date] [Author]

RE: Partial Rigidity (was ALGOR, EAGLE, ETABS, STAAD etc.)

[Subject Prev][Subject Next][Thread Prev][Thread Next]
>Based on the analysis result, we can design member
>sections, and then design the connector. It is necessary to calculate
>the connector stiffness and check if the connector stiffness is
>satisfied with analysis output. If not, adjust the connector strength
>and re-start the process again.
Not to rain on the parade, but I have a couple of problems with 'partial 
rigidity,' and I think they bear on the notion of an 'exact' analysis. 
First, connections aren't linear. Bolted connections in particular 
involve friction and slip into bearing; both welded and bolted 
connections carry residual stress (preload in the case of bolts) and both 
have a very complex stress distribution, despite the way we idealize the 
design. Consequently I doubt that connections really have a definite 
'stiffness' independent of fabrication and loading history. I think I 
could take a guess, maybe even do a 3D FEA model of my idealized 
connection, but I wouldn't bet the farm that it'd match the real world 
response in any recognizable way.

Statical determinacy also has a role, particularly when you're talking 
limit analysis. Indeterminate load components redistribute themselves; 
determinate loads don't, so it's hard to imagine the effects of guessed 
stiffnesses on the ultimate load assessment. Perhaps it's possible, but 
the underlying question is whether it can be done accurately (in the 
classical sense of matching performance in the field) and whether it 
makes economic sense.

Forgetting the effect of the idealizations for a moment, the process 
described by Luo seems open ended. At both ends. You need loads to assess 
strength, although the loads depend on the stiffness. But you don't know 
the stiffness until the connection is designed, which isn't possible 
without loading. There doesn't seem to be any place to start the process 
unless you've  started already. It's different for member selection, 
because you can select sections based on statical determinacy and tighten 
things up later on. When you're dealing with indeterminate loading 
stiffness starts to affect load transfer from the beginning--make a 
connection stiffer and it picks up more load, possibly requiring 
redesign, which again affects the stiffness. It has the faint odor of a 
diverging process.

It seems to me that we're adding an enormous amount of complexity to 
include second order effects when we really don't have all that firm a 
grip on all the first order approximations we use in connection design. I 
make a fairly good living doing finite element analysis; it's a 
marvellous tool for putting numbers on judgement calls. But it's real 
easy to get dazzled by the ability to do difficult arithmetic and assume 
you have just as much control over fabrication, loading and service 
response as over the computer display. 'Tain't so, McGee.

Christopher Wright P.E.    |"They couldn't hit an elephant from
chrisw(--nospam--at)skypoint.com        | this distance"   (last words of Gen.
___________________________| John Sedgwick, Spotsylvania 1864)
http://www.skypoint.com/~chrisw