The OpenSees Community

Thanks for your reply. I guess in the current state, OpenSees can not return the stiffness matrix before performing analysis. As a suggestion, it would be great to achieve the stiffness matrix before analysis. I am not sure if it is possible.

Hi

I would like to report a problem that I am not sure if it is a bug. I would like to obtain the stiffness matrix of a truss in openseespy. I use printA and FullGeneral system after ops.analysis. However, if the truss is unstable, it returns this error and quit the python console. It works properly if truss is stable. Returned error is:

FATAL FullGenLinSOE::getX - vectX == 0

Would you please help me out how to achieve a stiffness matrix for a non-internal stable truss?

Hi all.

I want to learn more about reliability analysis features of OpenSees. So, I started with the youtube video of https://www.youtube.com/watch?v=Nsv5otRuWLw. When I tried to run the examples using my standalone OpenSees with ver 2.5. I failed because I got this error: invalid command name "reliability"

I downloaded version 3 from GitHub and I compiled it myself on Ubuntu following the makefile instructions and I also modified the makefile in addition as follows,

RELIABILITY = YES_RELIABILITY &

in the section of MACHINE_NUMERICAL_LIBS, I ADDED: $(RELIABILITY_LIBRARY) \

and

ifeq ($(RELIABILITY), YES_RELIABILITY)
RELIABILITY_LIBRARY = $(HOME)/lib/libReliability.a
else
RELIABILITY_LIBRARY =
endif

but still, after compilation, I get the same error of: invalid command name "reliability"

Would you please help me out!

Best wishes,
Hirad

Hi,

Did you find out what is the problem regarding this error?

It got solved, there is a problem in support nodes and it's ambiguously mentioned in the literature.

Dear selimgunay, All units are consistent and in SI. I modified the code as you suggested and replaced the Iz 2.9666e-6 with 8.3333e-07, but no improvement in the result can be seen. The output of the code is as follows,
FullGenEigenSolver::solve() - the eigenvalue 9 is complex with magnitude 1.38843e+15
FullGenEigenSolver::solve() - the eigenvalue 10 is complex with magnitude 1.38843e+15
FullGenEigenSolver::solve() - the eigenvalue 13 is complex with magnitude 1.89559e+08
FullGenEigenSolver::solve() - the eigenvalue 14 is complex with magnitude 1.89559e+08
FullGenEigenSolver::solve() - the eigenvalue 16 is complex with magnitude 9.71949e+07
FullGenEigenSolver::solve() - the eigenvalue 17 is complex with magnitude 9.71949e+07
FullGenEigenSolver::solve() - the eigenvalue 21 is complex with magnitude 9.00654e+06
FullGenEigenSolver::solve() - the eigenvalue 22 is complex with magnitude 9.00654e+06
FullGenEigenSolver::solve() - the eigenvalue 25 is complex with magnitude 791442
FullGenEigenSolver::solve() - the eigenvalue 26 is complex with magnitude 791442
eigen values: -2.771764e+15 -1.388434e+15 -1.388434e+15

I tried to model a simply supported bridge, but I can't reproduce the results in the literature. The description of the truss is as follows,
"To demonstrate the flexibility of the approach, a simply supported bridge is optimized for its weight minimization with several cases of frequency constraints. Members on the lower chord are represented by beam elements with fixed rectangular cross sections B = 8 cm and H = 5 cm. Others are modeled as bar elements with initial sectional areas A = 1 cm2 . Young’s modulus is E = 2.1 × 1011 Pa, and the material density is ρ = 7800 kg/m3 for all elements. The initial configuration of the structure is shown in Fig. 6. A nonstructural mass m = 10 kg is attached at each of the nodes on the lower chord. The natural frequencies of the truss are about 20, 40 and 60" with some variations in the third or fourth floating point.
The schematic of the truss is depicted in : https://image.ibb.co/fkBQCa/Ground_Structure.png
--
First I modeled all elements as bar elements and it resulted in natural frequencies as "20.35 40.00 60.27". I believe that It came to a correct K matrix but the M matrix somehow was wrong. I surveyed the literature and I found out that the lower chord elements should be beam elements and I modified my model. But it still gives me very wrong Frequencies that are mentioned at the end. I also supposed that Iz in the beam element command of opensees should be (bh/12)*(b^2+h^2) and cosistent mass matrix as (rho*cross-section). I will appreciate if your help me to modify the model to repeat the literature results.

My tcl code is as follows:
model BasicBuilder -ndm 2 -ndf 3
node 1 0.0000 0.0000
node 2 1.0000 0.0000
node 3 1.0000 0.9392
node 4 2.0000 0.0000
node 5 2.0000 1.3270
node 6 3.0000 0.0000
node 7 3.0000 1.5063
node 8 4.0000 0.0000
node 9 4.0000 1.6086
node 10 5.0000 0.0000
node 11 5.0000 1.6679
node 12 6.0000 0.0000
node 13 6.0000 1.6086
node 14 7.0000 0.0000
node 15 7.0000 1.5063
node 16 8.0000 0.0000
node 17 8.0000 1.3270
node 18 9.0000 0.0000
node 19 9.0000 0.9392
node 20 10.0000 0.0000
fix 1 1 1 1
fix 20 1 1 1
uniaxialMaterial Elastic 1 2.1e+11
mass 1 10 10 0
mass 2 10 10 0
mass 4 10 10 0
mass 6 10 10 0
mass 8 10 10 0
mass 10 10 10 0
mass 12 10 10 0
mass 14 10 10 0
mass 16 10 10 0
mass 18 10 10 0
mass 20 10 10 0
geomTransf Linear 1
element elasticBeamColumn 1 1 2 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 2 1 3 0.0002983800 1 -rho 2.33 -cMass 1
element Truss 3 2 3 0.0001109800 1 -rho 0.87 -cMass 1
element elasticBeamColumn 4 2 4 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 5 3 4 0.0001009100 1 -rho 0.79 -cMass 1
element Truss 6 3 5 0.0002595500 1 -rho 2.02 -cMass 1
element Truss 7 4 5 0.0001261000 1 -rho 0.98 -cMass 1
element elasticBeamColumn 8 4 6 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 9 5 6 0.0001197500 1 -rho 0.93 -cMass 1
element Truss 10 5 7 0.0002426400 1 -rho 1.89 -cMass 1
element Truss 11 6 7 0.0001358800 1 -rho 1.06 -cMass 1
element elasticBeamColumn 12 6 8 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 13 7 8 0.0001477100 1 -rho 1.15 -cMass 1
element Truss 14 7 9 0.0002564800 1 -rho 2.00 -cMass 1
element Truss 15 8 9 0.0001129500 1 -rho 0.88 -cMass 1
element elasticBeamColumn 16 8 10 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 17 9 10 0.0001319900 1 -rho 1.03 -cMass 1
element Truss 18 9 11 0.0002921700 1 -rho 2.28 -cMass 1
element Truss 19 10 11 0.0001000400 1 -rho 0.78 -cMass 1
element elasticBeamColumn 20 10 12 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 21 10 13 0.0001319900 1 -rho 1.03 -cMass 1
element Truss 22 11 13 0.0002921700 1 -rho 2.28 -cMass 1
element Truss 23 12 13 0.0001129500 1 -rho 0.88 -cMass 1
element elasticBeamColumn 24 12 14 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 25 12 15 0.0001477100 1 -rho 1.15 -cMass 1
element Truss 26 13 15 0.0002564800 1 -rho 2.00 -cMass 1
element Truss 27 14 15 0.0001358800 1 -rho 1.06 -cMass 1
element elasticBeamColumn 28 14 16 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 29 14 17 0.0001197500 1 -rho 0.93 -cMass 1
element Truss 30 15 17 0.0002426400 1 -rho 1.89 -cMass 1
element Truss 31 16 17 0.0001261000 1 -rho 0.98 -cMass 1
element elasticBeamColumn 32 16 18 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 33 16 19 0.0001009100 1 -rho 0.79 -cMass 1
element Truss 34 17 19 0.0002595500 1 -rho 2.02 -cMass 1
element Truss 35 18 19 0.0001109800 1 -rho 0.87 -cMass 1
element elasticBeamColumn 36 18 20 0.0040000000 2.1e+11 2.9666e-6 1 -mass 31.20 -cmass 1
element Truss 37 19 20 0.0002983800 1 -rho 2.33 -cMass 1
puts "eigen values: [eigen -fullGenLapack 3]"

Result of the code:
FullGenEigenSolver::solve() - the eigenvalue 9 is complex with magnitude 2.41214e+15
FullGenEigenSolver::solve() - the eigenvalue 10 is complex with magnitude 2.41214e+15
FullGenEigenSolver::solve() - the eigenvalue 13 is complex with magnitude 1.90904e+08
FullGenEigenSolver::solve() - the eigenvalue 14 is complex with magnitude 1.90904e+08
FullGenEigenSolver::solve() - the eigenvalue 16 is complex with magnitude 9.83875e+07
FullGenEigenSolver::solve() - the eigenvalue 17 is complex with magnitude 9.83875e+07
FullGenEigenSolver::solve() - the eigenvalue 20 is complex with magnitude 8.9533e+06
FullGenEigenSolver::solve() - the eigenvalue 21 is complex with magnitude 8.9533e+06
FullGenEigenSolver::solve() - the eigenvalue 25 is complex with magnitude 842310
FullGenEigenSolver::solve() - the eigenvalue 26 is complex with magnitude 842310
eigen values: -3.007715e+15 -2.412142e+15 -2.412142e+15

Dear Frank,
I considered the consistent mass and it worked like a charm. Thanks for your incessant help and suggestions. I'm also grateful for this superb OpenSEES that you and your colleagues provided for scientific use freely.

Dear Frank,
First of all, thanks for your fruitful comment. I thought again and again about my problem and I'm now sure about that the mass matrix is not correct. An expert suggested me that my model looks ok but I have not yet added the lumped mass. He added that "this is done by declaring a mass element with the appropriate mass/inertial properties and then creating point mass elements where they are required." It's possible that If I consider additional mass the frequency reaches to 7.0000 but I don't know how can I implement the lumped mass to reach the frequencies reported in the literature. Now my results with considering mass on both DOFs on nodes are as follows,
omega^2=[2.134420e+03 1.191119e+04 1.704385e+04 1.802558e+04 3.568710e+04]
and frequency=[7.35 17.37 20.78 21.37 30.07].

I want to obtain the natural frequencies of a benchmark truss which is famously known as 10-bar planar truss. It consists of 10 truss members and 6 nodes. Two nodes are its supports and other four nodes are free nodes. The modulus elasticity is 6.89e10 and the added non-structural mass to the free nodes is 454 kg. Fig 1 shows the schematic of the truss and the obtained cross-section areas of this problem is listed in Fig 2. This figure also lists the natural frequencies obtained by different trusses. I tried to simulate the truss obtained by MC-TLBO which is listed in the last column but I faced a problem:

I didn’t reach the natural frequency in the literature and my model resulted in omega^2=1.575894e+04 and f=(omega/(2*pi))=19.98 Hz. It’s notable that I added some negligible massed on the other degree of freedom of the nodes according to OpenSEES official examples to obtain other natural frequencies.

The figures are referred from the [1] and The Tcl model is attached by myself (The units are in SI).

Any comment will be appreciated and thanks in advance for your time.

Figs.
Fig 1.
http://3.1m.yt/iCK85c-.png

Fig 2.
http://3.1m.yt/hxKpuE0.png

TCL model:
wipe
model BasicBuilder -ndm 2 -ndf 2
node 1 0.0000 0.0000
node 2 9.144 0.0000
node 3 18.288 0.0000
node 4 0.0000 9.144
node 5 9.144 9.144
node 6 18.288 9.144
fix 1 1 1
fix 4 1 1
uniaxialMaterial Elastic 1 6.89e+10
element Truss 1 4 5 35.8507e-4 1
element Truss 2 5 6 14.8424e-4 1
element Truss 3 1 2 35.5768e-4 1
element Truss 4 2 3 14.9305e-4 1
element Truss 5 2 5 0.645e-4 1
element Truss 6 3 6 4.6249e-4 1
element Truss 7 2 4 23.9816e-4 1
element Truss 8 1 5 24.2358e-4 1
element Truss 9 3 5 12.6977e-4 1
element Truss 10 2 6 12.3319e-4 1
mass 2 454 1e-9
mass 3 454 1e-9
mass 5 454 1e-9
mass 6 454 1e-9
puts [eigen 1]

Refs.
[1] Farshchin, M., C. V. Camp, and M. Maniat. "Multi-class teaching–learning-based optimization for truss design with frequency constraints." Engineering Structures 106 (2016): 355-369.

As to my knowledge and not practically, I know that first of all you should download the latest stable source code from the repository and compile it with the corresponding MAKEFILE for the version of linux. TCL libraries are essential too.
To download the source code from repository you should follow these instructions:
http://opensees.berkeley.edu/OpenSees/developer/svn.php
and to find the desired release tag you can find it out here:
http://opensees.berkeley.edu/OpenSees/changeLog.php
for building and compiling the source code and additional tcl libraries you should follow this:
http://opensees.berkeley.edu/OpenSees/d ... builds.php
and for MAKEFILE repository I guess these two sources are appropriate:
http://opensees.berkeley.edu/WebSVN/lis ... 61065d6b32
https://github.com/lge88/OpenSees/tree/master/MAKES

Hi, I want to model this 25 bar truss benchmark. the schematic of the benchmark and the loading conditions are attached.
How can I define this kind of loading in my code? Because there are multiple loading on nodes 1 and 2. Is it possible with multiple pattern plains? In a one sentence, How can I model this particular loading which is depicted in the picture?

[url]http://pasteboard.co/9rpVSLdj.png[/url]
[url]http://pasteboard.co/feezMfdv7.png[/url]

I tried in this way, but it doesn't work, since the deflection in the nodes should be lower than 0.35 inch.
Any help would be appreciated. THX in advance.

P.S. however it's mentioned that the second load on the node 2 on y-axis should be -2 in the picture, but the correct loading is -20 which I checked in other references.
[code]
wipe
model BasicBuilder -ndm 3 -ndf 3
node 1 -37.5000 0.0000 200.0000
node 2 37.5000 0.0000 200.0000
node 3 -37.5000 37.5000 100.0000
node 4 37.5000 37.5000 100.0000
node 5 37.5000 -37.5000 100.0000
node 6 -37.5000 -37.5000 100.0000
node 7 -100.0000 100.0000 0.0000
node 8 100.0000 100.0000 0.0000
node 9 100.0000 -100.0000 0.0000
node 10 -100.0000 -100.0000 0.0000
fix 7 1 1 1
fix 8 1 1 1
fix 9 1 1 1
fix 10 1 1 1
uniaxialMaterial Elastic 1 10000
element Truss 1 1 2 0.0140 1
element Truss 2 10 3 0.6740 1
element Truss 3 8 3 1.6470 1
element Truss 4 7 3 2.6900 1
element Truss 5 1 3 2.9670 1
element Truss 6 2 3 2.0120 1
element Truss 7 10 6 2.6900 1
element Truss 8 9 6 1.6470 1
element Truss 9 7 6 0.6740 1
element Truss 10 1 6 2.9670 1
element Truss 11 2 6 2.0120 1
element Truss 12 3 6 0.0110 1
element Truss 13 9 4 0.6740 1
element Truss 14 8 4 2.6900 1
element Truss 15 7 4 1.6470 1
element Truss 16 1 4 2.0120 1
element Truss 17 2 4 2.9670 1
element Truss 18 3 4 0.0110 1
element Truss 19 10 5 1.6470 1
element Truss 20 9 5 2.6900 1
element Truss 21 8 5 0.6740 1
element Truss 22 1 5 2.0120 1
element Truss 23 2 5 2.9670 1
element Truss 24 6 5 0.0110 1
element Truss 25 4 5 0.0110 1
timeSeries Linear 1
pattern Plain 1 1 {
load 1 1 10 -5
load 2 0 10 -5
load 3 .5 0 0
load 6 .5 0 0
load 1 0 20 -5
load 2 0 -20 -5
}
system BandGeneral
numberer RCM
constraints Plain
integrator LoadControl 1.0
algorithm Linear
recorder Node -file example.out -time -node 3 -dof 1 2 disp
analysis Static
logFile "log.txt"
analyze 1
print node
[/code]

First of all, thx for your reply. I'm repeatedly calling it in two sections; first I take stiffness matrix in a file with printA and I check the eigenvalues for stability and second, I run force/displacement analysis and it results in two files which include axial forces and displacement, so I guess I'm repeatedly calling it.
By the way, is there any way that I can calculate eigenvalues in tcl file without taking output and checking stability or for the second section that I need stresses in every element? I didn't find any command to do it directly in tcl.

correction : The code runs smoothly on Mac OS X but it crashes without any errors in Windows after hundreds of computations.

Hi, I have written a code to find optimal truss structure using evolutionary algorithms. Furthermore, these kind of algorithms are based on numerous computations. my code produces some text outputs and i read them from Matlab, these outputs are stiffness matrix,axial forces and displacements in nodes. The code runs smoothly on Mac OS X but it crashes without any errors in Windows, I have tested Windows Server 2012 and Windows 8, and Matlab 2013,2015 and OpenSees 32/64 bit. but this problem persists. I don't know what's wrong but I guess there is a problem with Windows in handling text files in linkage with Matlab. Any idea or insight would be appreciated to help me in solving this irritating problem.