Two improvement requests

pbeng · Post by **pbeng** » Thu Mar 12, 2009 3:09 am

Good Morning California

)

During the last workshop Frank mentioned that he wished a consultant would run with OpenSees on a project. I've done that on two isolated structures now, one in Ottawa (Canada) and one in Connecticut.

GUI issues were handled with a translator to FEMAP, the pre/post processor I use in other work.

I used zerolength elements for isolators, they worked well. The major shortcoming for my applications were that their directional characteristics are uncoupled. For example, if a material yields at 10 kips, and I apply it in the X and Z directions, its resultant in the XZ plane will be more than 10 kips. Is there any way to add the capability to couple directions?

Second improvement suggestion, a geometry query/echo tool. Geometry had to be double programmed so to speak, once to input, once to puts (the print -node method is limited in format flexibility). Would be nice if one could just write a loop with an OOP method such as get Nd.ID, then format a file with ND.ID Nd.x Nd.y Nd.z

I've written a lumped mass script for elastic beam elements, will post it soon. Would be a nice addition if that capability were added as well.

Tip of the hat in thanks to Frank and Silvia for making this excellent software accessible...

Cheers,

Paul Bradford

silvia · Post by **silvia** » Thu Mar 12, 2009 11:31 am

Paul,
have you looked at BuldingTcl? please do and let me know what you think.
i will be adding isolator models once i have it running well.
Also, why not the zeroLengthND Element?

silvia · Post by **silvia** » Thu Mar 12, 2009 11:31 am

or a zero-length section element?

pbeng · Post by **pbeng** » Sun Mar 15, 2009 4:48 am

Speed, simplicity, efficiency. The zerolength section approach I tried and got to work, but for reasons I can't recall rejected in favor of the zerolength uniaxial elements. The zerolengthND element; the desciption of the ND materials have the smell of complexity, a luxory I couldn't afford. In the context of a new and somewhat daunting fea package, simpler elements help.

The couple uniaxial material approach I think has a lot of promise. It's clear how the material behaves, debugging is easier, they run fast, a large libray can be built for various kinematic relationships. Some fea programs have a user input custom 1D material option, that'd be handy.

Also, a shout of thanks to the person who thought of the parallel material, having that capability helped tremendously.

Cheers...