The OpenSees Community

yes .. we can't really keep the graphics parts operational .. best record them and use matlab

go read my dissertation, ignore chapters on parallel programming

opensees.berkeley.edu/OpenSees/doc/fmkdiss.pdf

don't use a recorder at all .. get whatever quanaty you are recording from the command line and write them to a file.

how do results compare with Umfpack

no .. but just end all but 1 processor, e.g.

# do parallel stuff
if {$myID == 0} {
# do sequential stuff
}

the prupose of 1 analysis step is to find the next converged stp, given the current converged step. In the nonlinear analysis the model usually will go through a number of trial steps. these are not on the solution path, but trial solution points. If the convergence test is satisfied at a given trial step, the analysis step stops. before it returns, it tells the model that this is a converged step and hence the need for the 2 methods. if no converged step is found and the analysis fails, the revertTo LastCommit methods are invoked.

it works fine for me .. what version and operating system are you using?

what version of linux and what version of OpenSees did you build.

these are penalty factors that enforce the constraing .. it depends on the model you have. you want something large enough to enforce the constraint (as number goes towards infinity the constraint is enforced exactly) , but not too large to effect the conditioning of the global stiffness matrix and hence the results .. a number a few factors larger than the stiffness of the largest term in the matrix is usually sufficient

force, stiff, pressure, gap if 2d just force and stiff if 3d.

you might want to look at the source code for the element .. as Selim mentioned, the disp beam element assumes linear curvature and constant axial deformation along the length .. the equations are pretty standard for beams, they can be seen here:

http://opensees.berkeley.edu/wiki/image ... EvsDBE.pdf

what exactly are you expecting .. the time, given your model, is the load factor applied to enforce the constraint.

sorry for the delay .. can you try the latest

http://opensees.berkeley.edu/OpenSees/c ... Mac.tar.gz
http://opensees.berkeley.edu/OpenSees/c ... 64.exe.zip

that is a modelling choice i leave to you .. if you are a student you might want to find some expeimental data and try to see what best fits .. you might do both and see how the differences effect the results .. larger displacements, smaller forces if pure steel! .. in actuality probably between the 2 as the connection not perfect .. i encourage you to play with it.

what is the revision number that spits out after the version number