huangyhg

modeling eccentric connections
my first project as a rooky structural engineer has been to design a simple steel frame building. all of the beam-to-column connections are simple shear connections.
i am concerned that the eccentricity of loading for these columns may cause excessive moments to be induced in them. i'm using 5x5 square tubes with single plate shear connections. thus, the eccentricity of load is about 5.5" (bolt line to cl).
other engineers in my office do not concern themselves with the eccentricity of the load and do not treat the columns as beam - columns.
the potential for moment to be generated, especially if i consider the possibility of some bays being loaded while others are not (no counter balance), seems as though it should be investigated.
i calculate the moment induced by multiplying the load on the column by the 5.5" eccentricity. is there some other way to view this situation that makes it acceptable to ignore the moment due to eccentricity?
i've been reading up on torsion and i've learned that the torsion induced by side loading a member is not equal to the joist reaction x eccentricity of load. one has to apporach it from a deformation compatibility standpoint. maybe that could be the case here?
if anyone has any advice, it would be greatly appreciated.
you are on the correct path!
in the design of pressure vessels it is common to classify the stresses into two categories: primary stresses, those that are required to satisfy equilibrium, and secondary stresses, those, additional to the primaries, that are required to satisfy compatibility.
a point that is often not understood by designers, is that the classification into primaries and secondaries is up to the designer, and depends on how the structure is modeled.
to come back to your situation, if you decide to model the beam to column joint as pinned, then the bending moment at column top develops a secondary stress: in fact you might develop a plastic hinge there, but this of course doesn't mean that the structure will fail.
please note that i referenced a pinned connection, meaning that you will place the axis of rotation at the intersection of column and beam neutral axis, so that you have no moments in the column: that is the commonest choice, but it is up to you to decide where to place the pin.
a consequence of that choice is that you will design the connection to shear only, as there is a negligible moment there (though there is one).
if you make the coice of calculating the structure with a pin at the shear connection (5.5" to column cl), then you get primary moments in the column that you'll be required to account for, but you also get the benefit of a little shorter beam span!
above was the theory. of course you will have to follow the requirements of any applicable code. the point is that most of the existing codes for civil structures implicitly require the designer to consider only primary stresses: e.g. they allow for modeling a frame as a truss frame (pinned connections) whenever such behavior satisfies the designer (and this even if no actual truss frame is built with real pins in the connections!).
in modern times limit or plastic analysis allows for a better understanding and account of structural phenomena, but hopefully this will never be a requirement for simple structures.
prex
yes, you should account for connection eccentricity in your design. you should also consider live load skip loading patterns (if you're using ubc, see sections 1607.3.2 & 1607.4.2).
agreed. you should consider the connection eccentricity, then design the column for compression and bending. however, if your loads are really small, the additional moment induced will prove to be negligible.
adamp - i concur with the above recommendation to consider the eccentricities on the columns. if your beams are for floor loading, keep in mind that for most of your connections you will most likely have a beam on the other side of the column and your eccentricity is really that due to an imbalance in live loads only. most codes require consideration of alternative bay loading of live loads.
thanks alot for your help guys (and gals if applicable). i suspected that the eccentricity would have to be accounted for.
what still confuses me however, is that the aisc manual has a table for the design of tubes sections subject to purely axial load.
if a simple shear connection (about as close to pure axial as i am able to get) still produces appreciable moments, then what the heck are the tables for?
it's all the more frustrating because other people in my office can crank out columns at least 3 time as fast as i can when they ignore the bending and go straight from the tables.
i have yet more questions for those of you who mentioned codes and pattern loading:
1. ubc makes it sound like you need to consider pattern loads only when you are designing a continuous structure. is that the case? if eccentricity is causing moment, then the worst case for a simple column (considering the interaction eqns.) might be some form of pattern load.
2. the pattern loadings described in the codes seem to only address the planar situation. if a colum is picking up loads from 4 sides simultaneously, are designers expected to consider all the possible permutations of pattern loads in the two different directions?
again, thank you very much for your thoughtful responses to my previous question.
the axial load column tables are still useful because they can be used in the interaction unity equation for combined axial and bending. see aisc specification section h. as a simplistic example, if your axial load is 80% of the axial capacity, then the bending moment is limited to 20% of the flexural capacity.
as for pattern loadings, the word "continuity" in the ubc slightly confuses the issue. the intent of the code is that the structure should be designed for any realistic pattern of loaded and unloaded areas that could produce the worst case effect. one skip loading pattern is mentioned that is considered to satisfy this requirement. certainly, the pattern loading should be considered in both directions if applicable.
thanks taro. the bit about using the axial load values in the table for the interaction equation is very helpful.
adam, don't worry about how the other engineers in your office can whipp up design lickety split. often when you are first starting out you need to do everything from scratch, then as you get more farmilliar with the codes and develope a feel for the material you too can design quickly. right now you are doing the right thing. take this advice that was given to me "leave no stone unturned".

__________________
借用达朗贝尔的名言：前进吧，你会得到信心!
[url="http://www.dimcax.com"]几何尺寸与公差标准[/url]