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

RE: effective length factor for sway moment frames? K for frames with partially restrained connections?

[Subject Prev][Subject Next][Thread Prev][Thread Next]
In clarification of my previous answer, there are two seperate programs
types that are called FEA.  

The first are programs like RISA, STAAD, STRUDL, etc. that I consider
"stiffness analysis" programs.  These programs, use a matrix stiffness
analysis of one type or another and typically determine the results based
upon an elastic analysis.  Through the use of a P-delta analysis you could
massage out factors similar to k factors.  I think Scott had mentioned the
fact that most of these just pull their k-factor out of a table from AISC
(that's my understanding also).

The second are programs like ADINA or others that typically model using
meshes etc. Programs of this type can typically take effects such as local
buckling, initial imperfections, etc. into account which physically
determine the actual in-place K factors (kind of an oxymoron statement since
K-factors are approximations to adjust our imperfect equations to more
accurately model physical performance).  These programs also typically take
more time to model since the imperfections and etc. must be entered.  How do
you know what imperfections will exist in the field???  I never get advanced
notice for fork lifts running into columns or somebody going crazy with a
cutting torch.

Most of us use computer programs of some type to design our buildings,
bridges, etc...  To a computer, looking up a K-factor in a table is as quick
or quicker than trying to model inelastic or nonlinear behavior.  Ever had a
program tell you a member had 300" of deflection (typically because
something was entered wrong)...  even if entered correctly, could that
member really deflect 300" without breaking, yielding, etc. and how would
this affect the other results...

***
My point is that there are tools or methods that can deliver more *precise*
approximations.  But more *precise* numbers are not always more *accurate*.
***

As far as Scott's original question, it is my understanding that the use of
a p-delta (2nd order) analysis does not completely eliminate the need for
the use of K-factors (definitly room for interpretation).  I would tend to
agree with Mr. Geschwindner who responded in MSC.


Greg Effland, P.E.

-----Original Message-----
From: Waterman Drinkwater [mailto:Drinkwater(--nospam--at)EQUATION.COM]
Sent: Friday, March 30, 2001 11:32 AM
To: 'seaint(--nospam--at)seaint.org'
Subject: RE: effective length factor for sway moment frames? K for
frames wit h partially restrained connections?


We are not discussing which field of computing can have assumption. In fact,
all engineering computing has assumptions.

What kind of assumption is for FEA of K factors? It is the ordinary
structural theory. If you can accept those ordinary theory for daily
structural analysis, there is nothing wrong to apply the same theory to
analyze K factors. Those analytic results are much more reliable and
accurate than those from alignment charts.

FEA of K factors does not require any extra input data. The same set of data
for "ordinary structural analysis" is sufficient for analysis of K factors,
and you can get the output. The cost is much less than applying the
alignment charts. A pentium II 300MHZ can analyze 200 members' K in 1
minute. If applying alignment chart, how much time is necessary?





> ----------
> From: 	Effland, Greg[SMTP:geeffland(--nospam--at)butlermfg.com]
> Reply To: 	seaint(--nospam--at)seaint.org
> Sent: 	Friday, March 30, 2001 10:50 AM
> To: 	seaint(--nospam--at)seaint.org
> Subject: 	RE: effective length factor for sway moment frames?  K for
> frames wit h partially restrained connections?
> 
> I agree that FEA might be more accurate than the old alignment charts
> (actually the use of K factors in general is an assumption) but I have to
> agree with Charles on this one.  The results of any computer program are
> only as good as the input information, assumptions made within the
> program,
> and how accurate they are for that particular case.  To my knowledge all
> FEA
> programs are based upon assumptions, otherwise all physical testing would
> be
> unnecessary.  The point I think Charles was making is that this is a
> complex
> system in general... it is not like looking at bending on a flat plate.
> How
> many times has anyone analyzed a pinned base plate with more than 2 anchor
> bolts/rods?  Even with 2 anchor bolts this is not a true pinned
> connection...  Besides, the numbers are statisically massaged after
> physical
> testing to achieve a best fit line or curve... this always leaves some
> cases
> that don't exactly agree with the reported findings.  Again, I agree with
> Charles the resulting answer whether reported in 4 or 6 decimal places is
> only as good as the assumptions which are never perfect.  Key here is to
> make as good of assumptions as possible and go on... unless you want to
> take
> a year or more to design 1 frame.
> 
> my 2 cents worth anyway,
> Greg Effland, P.E.
> 
> 
> -----Original Message-----
> From: Waterman Drinkwater [mailto:Drinkwater(--nospam--at)EQUATION.COM]
> Sent: Friday, March 30, 2001 9:16 AM
> To: 'seaint(--nospam--at)seaint.org'
> Subject: RE: effective length factor for sway moment frames? K for
> frames wit h partially restrained connections?
> 
> 
> Theoretical results satisfy all the governing conditions. The theoretical
> K
> factor is theoretical exact. They are more reliable than the ones from
> alignment charts. Those alignment charts were introduced before computer.
> For the time being, almost every structural engineer has a computer for
> analysis. Why should we keep applying those inaccurate charts?
> 
> When computers were not available for most structural engineers, engineers
> used some approximate methods, for example, two cycles, to analyze frame
> structures. No one uses two cycles since computer is everywhere. All the
> results are theoretically analyzed on computers. FEA of K factors is a
> must.
> 
> 
> 
> > ----------
> > From: 	Charles Espenlaub[SMTP:Cespenlaub(--nospam--at)martinaia.com]
> > Reply To: 	seaint(--nospam--at)seaint.org
> > Sent: 	Friday, March 30, 2001 9:03 AM
> > To: 	seaint(--nospam--at)seaint.org
> > Subject: 	RE: effective length factor for sway moment frames?  K for
> > frames wit h partially restrained connections?
> > 
> > If it is a "theoretical" analysis, you may get "an" exact value.  But
> > since
> > theory is based on educated assumptions, the exact value you get may
> not,
> > and probably will not, be the "exact" value that occurs in nature.  Just
> > because a computer can give you a four digit answer, does not mean,
> > necessarily, that it is the right answer.
> > 
> > 
> > Charles F. Espenlaub, III, P.E.
> > Martin-Espenlaub Engineering
> > Tel  215-665-8570
> > Fax 215-561-5064
> > 
> > 
> > 
> > -----Original Message-----
> > From: Waterman Drinkwater [mailto:Drinkwater(--nospam--at)EQUATION.COM]
> > Sent: Thursday, March 29, 2001 4:55 PM
> > To: 'seaint(--nospam--at)seaint.org'
> > Subject: RE: effective length factor for sway moment frames? K for
> > frames wit h partially restrained connections?
> > 
> > 
> > Try theoretical (FE) analysis. You may get the exact value.
> > 
> > WD