effective buckling length

huangyhg

effective buckling length
dear colleagues,
in order to take column buckling lengths equal to unity and use aisc-asd design procedures, can we use pdelta analysis taking notional forces into account arising from imperfections and shall notional forces be included in pdelta combination in conjunction with dead and live loads. if buckling lengths are taken to unity with an appropriate analysis, aisc requires max kl/r as 200 for columns, how can this be checked as taking k=1, or is this requiremet no longer valid.
regards
r.akbaba
k is a measure of end restraints of the   
note that k is also related to whether the column is in a sway frame or non-sway frame.  this is related to pdelta effects indirectly.  when pd effects are included in a second order analysis, we typically continue to use the k associated with the end constraints and the sway/non-sway condition.
also...one more little tid-bit.  aisc does not require kl/r to be less than 200.  the specification states that kl/r "preferably" should not exceed 200.  you can exceed 200, and the design equations are still valid.  see aisc commentary on b7 for a discussion on this.
please keep in mind that eurocode 3 clause 5.2.6.2 permits in-plane buckling lengths for the non-sway mode when second-order analysis is performed. pls. also refer to ssrc guide(3rd edition) which supports the same method, but the ec3 takes imperfection forces into account while ssrc does not.so taking k=1 with an appropriate second order analysis and imperfection forces, is an conservative approach due to ec3. my question is directly related to an adaptation to aisc/asd.
chandr (visitor)3 mar 01 11:40
asd chk: fa/fa + b.fb/fb =1, b being the magnifier due to pd.  in my opinion, you can use results from pd analysis for fa, b.fb.  but you still have to get a resonable value for k in calculating fa.  in such a case, you should do elastic buckling analysis to get k.  then, from the buckled shape you may reshape your model with initial imperfections and do pd or nonlinear analysis, ie. you have to have many nodes on one column.