ension only members in software packages

huangyhg

tension only members in software packages
hi,
i've found that some software packages have an option of specifying a particular member as a "tension only" member.
however,i've never used this functionality in my practise so far (though i am relatively new to the field!).
i am curious to know what is the advantage of specifying a member as a "tension only" member?
also,i found that this functionality of "tension only" member is found in the non-linearity option corresponding to structural analysis in some packages.
could anyone shed some light on this functionality of tension only members and cite some situations wherein you have used this option and in what context?
thanks in advance.
i have used tension only members in the analysis of steel frames where the cross bracing is steel flats. the steel flats are good in tension but next to useless in compression. hence specifying them as tension only   
bascially, after performing the analysis, the software you are using will search to see whether specified tension only   
also useful for modeling cable components.....
ed.r.
...and good for modeling contact--just put the tension-only member in on the opposite side of the member.
just be sure to run lots of simple test cases for which you know the answer.  i've tried tension-only   
"also,i found that this functionality of "tension only" member is found in the non-linearity option corresponding to structural analysis in some packages."
using tension-only members is by definition non-linear, because the structural stiffness matrix must be modified whenever a tension-only member is added or deleted from the structural model, and an iterative analysis is required until a converged solution is obtained for all load cases. if you are running several load cases, it is very possible that some tension-only members will be in tension for some load cases (e.g. wind from north, say), but compression for others (e.g. wind from south), so each such load case will conceivably have a different set of active members, each of which requires its own solution.
if you specify tension-only members, but run a linear analysis, most software packages would include all tension-only   
hi  all,
in an analysis, if a member is specified as tension-only,but during first analysis found in compression,then (as said by asixth) in the reanalysis(second time) that member would be deleted from the calculation and reanalysis carried out.this would thus modify the structural stiffness matrix (as pointed out by julianhardy).
but,julianhardy,i do not get your point when you say that-
"if you specify tension-only members, but run a linear analysis, most software packages would include all tension-only members in all load cases, regardless of whether they are in tension or compression"
can you kindly shed some light on this statement?
in any case even in linear analysis each load case would require its own solution and when the user generates combinations he could superimpose the laod cases to get the effects for combinations.right?>
i am trying to say that superposition is still valid even in tension only   
some packages will only calculate one stiffness matrix for one freedom case for all load cases.
so if an element defined as "tension-only" goes into compression for any load case than that element will be removed from the analysis. this isn't ideal and the developer producing the software should be notified.
risa-3d has a pretty good option for tension-only members.  they call it an euler buckling member.  these members can take compression up to their critical buckling load.  after the compression exceeds this value, it shuts it off and eliminates it from the stiffness matrix.  this helps with model convergence.  
sometimes with true tension-only members you can get a situation like asixth was discussing where the model will not converge because when the model shuts the member off and re-runs the analysis, the would-be strain in the member is a tensile one and the program will turn the member back on again.  next run off, next run on, and the solution will never converge.  if you allow the member to take some small compressive force the model will be able to converge - hence the euler buckling   
one reason to use tension on only is to avoid the seismic compactness criteria of aisc 341-05 seismic provisions section 14.2 for ocbf's.
i use both tension only and compression only elements to mimic chains/cables and foundation supports, respectively.  seems to work nicely (in cadre).