The OpenSees Community

you have to build a debug version of OpenSees in VS as we do not release a debug version .. running the released version will not work as there is no data in the exe with which the debugger can use.

to what end are you storing the state! .. the only alternative would be to recompute from the start.

assuming you have nodes at the column line floor intersections.
1) add 2 duplicate nodes for the ele ends.
2) use equalDOF to connect nodes to existing nodes for the translational dof
3) add elastic beam between these 2 new nodes
4) add a bilin material for moment-rotation of the spring
5) add 2 spring elements connecting the original nodes with 2 new nodes

a 4 node example should only have 1 element! .. the sript is running for me and produces the following when i do a print on 2 ele:

DKGQ Non-Locking Four Node Shell
Element Number: 653
Node 1 : 523
Node 2 : 524
Node 3 : 525
Node 4 : 526
Material Information :


DKGT Non-Locking Three Node Shell
Element Number: 690
Node 1 : 468
Node 2 : 479
Node 3 : 467
Material Information :

the dt of the analysis and the steps in the Path times series do not have to be the same .. OpenSees just uses linear interpolation to get the timeseries value at whatever time the domain is at.

a file editor such as notepad on windows.

when you start OpenSees type "pwd" to see what directory the application starts in .. use the cd command to get to the directory the file is in before you source it.

you have 0 unknown dof when you use the transformation method .. you might try the Penalty method

if you are getting results, then you can .. whether the results are correct or not is something else .. looking at the code in the Maxwell model, i would doubt it .. this of course assumes young Prof. Dimitrios Lignos got it correct .. perhaps you should send him an email. He is now is Switzerland.

int
Maxwell::setTrialStrain(double strain, double strainRate)
{
// Set Total Strain
Tstrain = Cstrain;
Tstress = Cstress;
Ttangent = Ctangent;

// Determine change in strain from last converged state
double dStrain = strain - Cstrain;

// Constant based on Viscous Damper Viscocity C and Axial Stiffness K
double tR = (C/pow(L,Alpha))/K;

// Integration Constants as a function of incremental time dt
// Note: The integration time dt = ops_Dt is coming in from the OPS_Globals
double A = Tstress * ( exp(-ops_Dt/tR) - 1.0);
double B = (K/2.0) * (1.0 + exp(-ops_Dt/tR) );

// Incremental Stress Calculation (considers the history of strain in A and B
double Dstress = A + B * dStrain;

// Total Stress Calculation (Tstress) is increment Stress Plus Previous Total Stress
Tstress = Dstress + Tstress;
Tstrain = strain;

// double DTangent = Tstress/Tstrain;

return 0;
}

the material you are using: Concrete07 was not added to the ObjectBroker code, i have added the code. if you can build the beast that is OpenSees you can start using the code know.

simple script for you to play with:

model Basic -ndm 2
set accel " Path 1 -dt 0.01 -filePath a.dat"
pattern UniformExcitation 1 1 -accel $accel

you should play with to sort out your problem, e.g.
if a.dat does not exist or is an empty file:

OpenSees tmp.tcl


OpenSees -- Open System For Earthquake Engineering Simulation
Pacific Earthquake Engineering Research Center
Version 3.0.0a (rev 6692) 32-Bit

(c) Copyright 1999-2016 The Regents of the University of California
All Rights Reserved
(Copyright and Disclaimer @ http://www.berkeley.edu/OpenSees/copyright.html)


WARNING - PathSeries::PathSeries() - could not open file a.dat
WARNING -- PathSeries::getDuration() on empty Vector


if a.dat exists and contains at least a space

tmp >OpenSees tmp.tcl


OpenSees -- Open System For Earthquake Engineering Simulation
Pacific Earthquake Engineering Research Center
Version 3.0.0a (rev 6692) 32-Bit

(c) Copyright 1999-2016 The Regents of the University of California
All Rights Reserved
(Copyright and Disclaimer @ http://www.berkeley.edu/OpenSees/copyright.html)


WARNING -- PathSeries::getDuration() on empty Vector

the only options are to figure out the integration weights (you might trying using the default ones and addinng other ones at points wanted with a 0 weight .. check if it works by doing it and comparing results .. they should not change) .. other option is to use disp based elements and put nodes at points wanted & use gauss lobatto.

not seeing the error:

fgrep ShellDKGQ ShellDKGT.*
ShellDKGT.cpp: s << "EL_ShellDKGQ\t" << eleTag << "\t";

can you post a 4 node example that just creates 4 nodes, 1 material and 1 element

the trick was only needed if there were no accelerations at the base. as you are using the Penalty method, the base nodes appear in the equations and Newmark is solving for them and updating accelarations at the node. Sometimes people when using Penalty, also keep the fixity in the dof which results in the Penalty method only imposing half the displacement.

Anyway back to your original problem: it is really related to the Newmark method and the assumption it makes on how the acceleration changes between time steps which is effecting the results you are getting. You can try cutting the time step until you get an answer at the base that you are happy with.

the accel should be 0 (unless you used Penalty) .. you should be able to set the accel by including the -accel acelSeriesTag to the time groundMotion (possibly an undocumented feature) .. if you used Penalty tou could check the results for the accel by creating a Matlab script and using the relations for changes to accel given changes to disp behind the Newmark algo.

you cannot use with disp beam element, your only choice is to switch to forcebeam element .. or put element in parallel with a force beam element with negligably small elastic section (can't be too small as has to invert it) and the shear aggregator you are using.