validation of Opensees with another software

[sreedharrao144]

Sir,
I modelled water tank with 4 noded shell elements and pinned at the bottom in 3D in Staad Pro. I modelled the same water tank with ShellMITC4 element and ElasticMembranePlateSection as section in opensees in 3D. All the properties like thickness,young's modulus, water pressure (water pressure is distributed at nodes and given as loads in opensees) are same in both cases. But I am not getting equal results. To assign pinned condition I used the following command. Is that correct if we want to use pinned boundary. I used the below one got the moments in 10 ^-12 or 10^-14 range. Those are very less so we can take it as zero. But the displacements and forces I am not getting equal. I got max error 34.0279 % ,20.91833333% and 17.84977778% at some nodes. I model has 89 nodes arround 15 nodes I am getting huge variation like above said values? Why this was occurred? It is due boundary conditions? Please reply me. can you provide any validated results or examples with another software. Because in organizations clients are asking validation.Please help me in this regard. I could not co relate the results. and also help me in assigning pinned boundary in my case.

Boundaries are assigned as follows (bottom)
fix 1 1 1 1 0 0 0
fix 6 1 1 1 0 0 0
................

[vesna]

I do not have any comparison of OpenSees model using shell elements with any other software.

Most of the software can account for gravity automatically witch OpenSees can not and most of the time this leads to different results. Try to apply gravity to your tank and see if you are going to get any better results.

As of boundary conditions I would fix torsional rotation as well.

[sreedharrao144]

Thank u vesna....thank u very much but the problem with clients is that they are asking validation because people are not aware opensees. Opensees is not graphical interface to understand easily.Basically opensees will be used by researchers and not popular in organizations.I used Opensees in my thesis work in masters.But in the organizations they are asking validation.I believe methodology will be different in different softwares. So results we can not get easily same from different softwares. Boundary conditions,compartability of problem to the software,behaviour of elements used,analysis precision etc there are actually several issues.Even small change in the matrix will tends to ill condition system of equation.In that case impossible to converge.Pivoting helps upto some extent.If model is big then that will be a challenge if we are using opensees.I faced lots of problems in opensees now got the confidence but interms of validation I am failed.I could not validate atleast simple problem. Any way thanks for the reply I will apply mass at each node and see the results. or I will apply vertically downward load at each node. I will let you know shortly.Please try to incorporate some worked out examples with other software vs opensees.So we can able to convey that opensees works well and can be used for all the problems.It is my request. If we can show the results with less than 10% error then that will be acceptable to all. So please work on in this regard of course I will also help in this issue in validation especially.

[sreedharrao144]

Vesna..... can you give me ur Gmail ID. I want to have some discussion on opensees. Actually in a command fix what actually it is taking about?

This command is used to construct single-point homogeneous boundary constraints.

fix $nodeTag (ndf $constrValues)

--------------------------------------------------------------------------------
$nodeTag integer tag identifying the node to be constrained
$constrValues ndf constraint values (0 or 1) corresponding to the ndf degrees-of-freedom.
0 unconstrained (or free)
1 constrained (or fixed)

EXAMPLE:

fix 1 1 1 1 1 1 1; # node 1: fully fixed

fix 2 0 1 0 0 1 0 # node 2: homogeneous constraints (movement = 0) at dof 2 and 5 only.


Code Developed by: fmk

Retrieved from "http://opensees.berkeley.edu/wiki/index.php/Fix_command"
if I want assign pinned boundary condition in 3D what will be the command. The results I should get at the pinned nodes displacements and moments,torison will be zero.How can I assign? The example is showning for fully fixed and what about the hinged boundary.The dof indicating here is in global directions are local directions of nodes. Means axial,bending,torsional directions in local or global X,Y and Z- directions irrespective of local. Please reply me.Please make me understand about this command. Justify the difference between movement and zero moment at pinned support.

[vesna]

With fix command you can restrain displacements or rotations of a node in any degree of freedom.

[sreedharrao144]

Thank U vesna. thank u very much.If we give 1 then that will restrain the node. For example
fix 2 1 1 1 1 1 (means node 2 have no displacement and no rotation in all X,Y,Z directions i.e. global X,Y,and Z directions.In this situation the node will have reactions means forces and moments in X,Y,Z i.e.in global directions.we can take it as fixed support.
fix 2 1 1 1 0 0 0 (means node 2 have no displacements in global X,Y and Z directions and has rotations in X,Y and Z global directions.In this situation the nodes have reactions in X,Y and Z global directions,rotations are allowed so there is moments about X,Y and Z directions means zero moments about x,y and z global directions.This supoort we can take it as pinned or hinged support.
fix 2 1 0 1 0 1 0 (in this case no displacements in x and z directions and no rotation about global Y direction.
Vesna please correct me if my understanding is wrong.Or above said fixed and pinned boundaries in 3D are correct or wrong. Please correct me if
I m thinking in a wrong way because I using opensees from 1.5 years but still in validation I am failed.
I tried the water tank problem without selfweight in STAAD Pro and opensees but could not get results.Still same percetage of error. Why this kind dicrepency in the results vesna. Please reply me soon.If u have any validated results with another software please inform me.

[sreedharrao144]

Dear vesna .....I need a favour from you.I am validating openses with Staad Pro.The problem I am dealing is a simple static analysis of a truss.But while anallyzing in opensees it is becoming trivial.Can u check my script and can u update the script to run.I used all the solvers available in opensees.But results I m getting is QNAN,IND.Please allot your few minutes on my problem. I am expecting your cooperation in validation of opensees with STAAD PRO.

script as follows:

#units :N and m
model basic -ndm 2 -ndf 2
file mkdir data;

node 1 0 0
node 2 9 0
node 3 0 6
node 4 3 6
node 5 6 6
node 6 9 6
node 7 0 10.5
node 8 9 10.5
node 9 2.25 10.5
node 10 6.75 10.5
node 11 4.5 10.5
node 12 1.5 11.42
node 13 7.5 11.42
node 14 3 12.34
node 15 6 12.34
node 16 4.5 13.26

set A [expr 0.2*0.4]
set E 2050E+09
set Iz 2.66E-04

section Elastic 1 $E $A $Iz

element truss 1 1 3 1
element truss 2 3 7 1
element truss 3 2 6 1
element truss 4 6 8 1
element truss 5 3 4 1
element truss 6 4 5 1
element truss 7 5 6 1
element truss 8 7 12 1
element truss 9 12 14 1
element truss 10 14 16 1
element truss 11 15 16 1
element truss 12 13 15 1
element truss 13 8 13 1
element truss 14 9 12 1
element truss 15 9 14 1
element truss 16 11 14 1
element truss 17 11 15 1
element truss 18 10 15 1
element truss 19 10 13 1
element truss 20 7 9 1
element truss 21 9 11 1
element truss 22 10 11 1
element truss 23 8 10 1

#suppots fixed

fix 1 1 1
fix 2 1 1

pattern Plain 1 Linear {
load 4 0 -65000
load 5 0 -65000
load 11 0 -15600
}

recorder Node -file data/disp1.out -time -node 1 -dof 1 2 disp;
recorder Node -file data/disp2.out -time -node 2 -dof 1 2 disp;
recorder Node -file data/disp3.out -time -node 3 -dof 1 2 disp;
recorder Node -file data/disp4.out -time -node 4 -dof 1 2 disp;
recorder Node -file data/disp5.out -time -node 5 -dof 1 2 disp;
recorder Node -file data/disp6.out -time -node 6 -dof 1 2 disp;
recorder Node -file data/disp7.out -time -node 7 -dof 1 2 disp;
recorder Node -file data/disp8.out -time -node 8 -dof 1 2 disp;
recorder Node -file data/disp9.out -time -node 9 -dof 1 2 disp;
recorder Node -file data/disp10.out -time -node 10 -dof 1 2 disp;
recorder Node -file data/disp11.out -time -node 11 -dof 1 2 disp;
recorder Node -file data/disp12.out -time -node 12 -dof 1 2 disp;
recorder Node -file data/disp13.out -time -node 13 -dof 1 2 disp;
recorder Node -file data/disp14.out -time -node 14 -dof 1 2 disp;
recorder Node -file data/disp15.out -time -node 15 -dof 1 2 disp;
recorder Node -file data/disp16.out -time -node 16 -dof 1 2 disp;

constraints Plain
numberer RCM
system SparseSYM
integrator LoadControl 1.0 1 1.0 1
algorithm Linear
# create the analysis object
analysis Static
# perform the analysis
analyze 1


Please update my script if I am wrong and I tried all the solvers but could not get results.it is a simple static analysis and elements also simple.Why this kind of problems in opensees?
Thanks and Regards,
Sreedhar Rao Gentela

[vesna]

I tested your file with OpenSees 2.3.0 and did not have any problems. Make sure to have "wipe" command at the beginning of the file.

Your understanding of "fix" command is correct.

I do not have idea what can be reason for not achieving the same results with STAAD Pro and OpenSees. Make sure that elements and materials that you use in the two software are the same.

[sreedharrao144]

Thank you vesna...but still I m getting nonsense results like QNAN and IND.If I keep wipe command at the top of the file then also same thing.I checked with opensees 2.3.1 and 2.3.0 versions.I could not get the results.I dealt big domain problem i.e soil structure interaction analyzed in opensees and got the results after facing so many problems.But in this small 2D problem I am facing problem. Please check once.I am using window7 proffessinal OS.processor core i5 and ram 4GB. Is that the problem for not getting numbers in output file? Opensees 2.3.1 and 2.3.0 are compartable with which OS. Please reply me soon vesna.Your help is really needed for me.That is why I asked you Gmail ID.So that I can contact you for all queries what I have in opensees. Because my career is depends on opensees scrpting.I am from structural engineering stream in my masters so please reply me soon. Shall I uninstall opensees and install again.

[sreedharrao144]

Dear vesna..sorry for disturbing you again and again.My situation is like that please understand.I m using TCL editor version 0.3.1 and opensees 2.3.1 version.Are those versions are compartible to each other or which version of Tcl editor I should use if I want to run opensees 2.3.1 effectively. From where I can download tcl editor which is compartible to opensees 2.3.1 version. Please reply to this mail. Are u getting proper results with those solvers specified in the script instead of QNAN and IND. In which platform u ran the program? I m using windows7 professional.

[sreedharrao144]

Dear vesna....one more clarification needed.In the problem I specified in model builder -ndm 2 -ndf 2 that means it is 2D problem and at each node 2 translations in x- and y- directions are possible no matter of rotation.
If -ndm 2 -ndf 3 means it is 2D problem and at each 2 translations in x and y directions and rotation about z is possible.
If -ndm 3 -ndf 4 means it is 3D problem and at each node 3 translations in x-,y-z- dierctions and rotaion about z is possible.
If -ndm 3 -ndf 5 means it is 3D problem and at each node 3 translations in x-,y-z- dierctions and rotaion about z,x is possible.
If -ndm 3 -ndf 6 means it is 3D problem and at each node 3 translations in x-,y-z- dierctions and rotaion about z,x,y is possible. Is it correct?
one more clarification in the above said truss problem I am getting QNAN and IND.I dont know how u analyzed. But the problem it is looking mixture of truss as well frame( i.e.linear beam column elemnts). In that case I need to define 3 DOF and all the elements I have to model as linear beam column element with all transformations that actually I am facing problem in transformation. You please post results of the nodal displacements next time.I am expecting your help in the validation and one more problem of water with shell elements in3D also going to validate with opensees.But atleast this simple problem first I am facing in validation in terms of analysis.How to deal this truss problem first? I could not get values in displacements and only sparseSYM is working in this case.Why this kind of complexity in opensees when we are dealing simple 2D problem. I need your help in execution of this problem......Please vesna help me. And thanks in advance for your updated file and result file.

[vesna]

Last time I ran your file and did not get any warning messages, so I thought it was good. This time I looked at the output files and I saw QNAN. Then I changed the system to band general and realized that your model has stability issues. Than I looked at your structural system (you can do it too by using display command: http://opensees.berkeley.edu/wiki/index ... ode_shapes) and saw that you were using truss elements (that have hinges at both ends) instead of beam-column elements for beam and columns. Shortly, your structural system is unstable and that is why you could not solve it (it is not related to windows, tcl editor, etc).

Dimension of the problem and degrees of freedom depend on the element. For example, truss elements and quadrilateral elements can be modeled with -ndm 2 -ndf 2, beam-column elements can be modeled with -ndm 2 -ndf 3 and -ndm 3 -ndf 6, quadrilateral elements can be modeled with -ndm 2 -ndf 2, -ndm 3 -ndf 3 and -ndm 3 -ndf 6 (depending on the element). However, I do not know of any element that has -ndm 3 -ndf 4 and -ndm 3 -ndf 5.

For -ndm 2 -ndf 2, the only degrees of freedom are translations in x and y.
For -ndm 2 -ndf 3, there are two translations (x and y) and rotation about z as degrees of freedom.
For -ndm 3 -ndf 3, the only degrees of freedom are translations in x, y, and z.
For -ndm 3 -ndf 6, the degrees of freedom are three translations (x, y, and z) and three rotations about x, y, and z.

[sreedharrao144]

Thank you vesna...thank you for the reply.I wondered initially how you got the results. But now you also came to know my problem.it was not showing but results are nonsense.So I tried to analyze the problem with all linear beam column elements (inclined,vertical and horizontal) but I am facing problem in transformations.I was showing error that wrong transformation. Can you provide any literature for transformation of inclined,vertical,and horizontal members.I have seen in the command manual but it was not working in my case.There is no example on assigning jntOffset $dXi $dYi $dXj $dYj.Please help me in my case if you can.Actually validation is very important in my case if I could not validate I am allowed to work on opensees.so I am asking vesna.Please help me in running this problem.Give some literature on transformation.I think command manual is insufficient to explain properly.Need some examples with inclined,vertical,horizontal in specifying $dXi $dYi $dXj $dYj.....Thank you again for your immmediate response for my queries.

[sreedharrao144]

Dear vesna ....can you explain the concept of geometric transformation (all the three options) with an example which includes numerical values. I am facing a big problem with geometric transformations with members.Please consider a general member (inclined) for explanation.Thanks in advance for your reply

[vesna]

I'm sorry but my role is not to provide solutions for OpenSees users but to guide you on how to get there. The idea of OpenSees is that everybody posts examples and that community members help each other.

Let me respond to your questions.

For a two-dimensional problem:
geomTransf Linear $transfTag <-jntOffset $dXi $dYi $dXj $dYj>

where -jntOffset $dXi $dYi $dXj $dYj is optional. By defining joint offset you make your ends rigid.

Linear transformation does not account for nonlinear geometry. Corotational transformation accounts for nonlinear geometry exactly and Pdelta accounts for nonlinear geometry by imposing certain approximations. You can read more about this in: "Earthquake Engineering: From Engineering Seismology to Performance-Based Engineering" by Bozorgnia and Bertero.

Example of inclined member in X-Y plane with an angle of 45 degrees and rigid ends with length of sqrt(2). Joint offset will be defined in the following way:
-jntOffset 1.0 1.0 -1.0 -1.0

Please read again command manual (http://opensees.berkeley.edu/wiki/index ... sformation) and you will understand this definition of joint offset.