building complex models

huangyhg

building complex models
folks,
if you are building a rather complicated structure like a stadium or a odd shaped twisting building or some such, how do you build the model in the analysis program?
1. do you start the model in the analysis software?
or
2. create it in a third party software and then import it into the analysis package?
i am looking to find information on how people do it. can programs like tekla be used to build structural models?
i used to use patran to build abaqus models in school. do such things exist in the industry?
check out our whitepaper library.
early versions of staad had pretty poor gui tools for constructing complex models.  i've had several projects where i drew the geometry in 3d cad and had staad pick up the dxf from there.
since the analysis programs' primary function is analysis, they usually come up short in the tools for building complex geometries.
if you "heard" it on the internet, it's guilty until proven innocent. - dcs
staad has lagged the rest of the industry when it comes to the ease of the user interface. we have guys here who still edit the text batch file by hand. they seem to revel in the difficulty of it - maybe it's a job preservation thing.
risa is excellent for building models. it has always prided itself on its superior graphical interface but there still are limits. if you are looking at a host complex curves it may be more sensible to use a cad platform to create the geometry.
i've had cad geometry sent to me that looked ok but was full of gaps/slivers etc. that was impossible to fix. in my view it's quicker and easier to build the geometry in the software itself than have to repair cad gemoetry which was produced for a different purpose. using abaqus cae i find it's fairly easy and quick to build the geometry i need.   
corus
hi,
if you have the model in tekla then it is possible to import the geometry in staad.pro. i doubt it is possible with other fem programs and definitely not with abaqus.
i used patran long time ago in my studies to generate geometry of fem models. it wasn't so pleasant. however i do not know how it is now, perhaps it is better as is the case with other commercial softwares.
since this seems to be an academic exercise with fixed resources (patran&abaqus), the only way i see is to generate the geometry in patran and run in abaqus. important thing is to focus on learning fem modelling, interpretation of analysis results and understanding structural behaviour and its relation to your modelling assumptions etc.
goodluck!