Hello everyone,
I need to extract the mode shape vectors of some cantilever plate to make a correlation between some of them analytically.
I used the workbench to simulate the problem and added the next APDL command to extract both mass and stiffness matrices in MMF:
/AUX2
COMBINE, FULL
/POST1
*SMAT, MatKS, D, IMPORT, FULL, file.full, STIFF
*SMAT, MatMS, D, IMPORT, FULL, file.full, MASS
*Export, MatKS, MMF, matK_MMF.txt
*Export, MatMS, MMF, matM_MMF.txt
Then I did modal analysis , and I got the two files of mass and stiffness matrices in MMF .
I used the next Matlab code to solve the eigen problem in order to extract mode shapes.
clc;
clear all;
format shortG;
format loose;
load matK_MMF.txt;
K = zeros(462,462);
for r = 2:5515
K(matK_MMF(r,1), matK_MMF(r,2)) = matK_MMF(r,3);
end
disp (K)
load matM_MMF.txt;
M = zeros(462,462);
for r = 2:1999
M(matM_MMF(r,1), matM_MMF(r,2)) = matM_MMF(r,3);
end
disp(M)
cheq=linsolve(M,K)
[Mode,Lamda]=eig(cheq);
lamda=diag(sort(diag(Lamda),'ascend')); % make diagonal matrix out of sorted diagonal values of input 'Lamda'
[c, ind]=sort(diag(Lamda),'ascend'); % store the indices of which columns the sorted eigenvalues come from 'lamda'
omegarad=sqrt(lamda);
omegaHz=omegarad/pi/2
mode=Mode(:,ind)
This code ran syntactically without any errors .
I checked for the first natural frequency, omegaHz(1,1), which is supposed to be 208.43 Hz as shown in the workbench analysis , but unfortunately it was 64023 Hz .
Would you please show me what is wrong with that problem?
Or is there any possible way to extract the mode shape vectors or modal matrix directly from ANSYS ?
Regards
Mohamed A M Salem Please find enclosed the normal modes in ASCII (.post) and binary format (.hdf). You can use any hdf viewer to open .hdf files, e.g.
https://www.hdfgroup.org/downloads/hdfview/
https://github.com/marts/hdf5view
Nils Wagner appreciated for your concern, but I think some misunderstanding happened. I need a methodology to extract the Eigen vectors of any structure simulated by ANSYS
regards
Mohamed, I think you must be more specific regarding what is it that you want to achieve, and define what you mean by "extracting". Do you aim at obtainaing numerical values that you would operate on within ANSYS APDL capabilities, or do you want to export the numerical values to work on them outside of ANSYS? ANSYS has basic modal correlation capabilities (MAC estimation for example, see RSTMAC command).
Dear Mohamed,
try the command lines provided in the attached file. They work for a full 3D model, but you can easily modify them for other kind of elements.
Best regards,
Marco Montemurro
Mohamed A M Salem The code provided by Marco Montemurro is the one you need.
Hello Marco Montemurro
appreciated for your valuable contribution, I just need to discuss two issues:
1st issue >> I noticed that you mainly depend on displacements along the axes (X,Y,Z) to define eigen vectors where the used polynomials of the ANSYS are included ( they are displacements not eigen vector extracted from Mass, Stiffness eigen problem) also the rotations about the 3 axes are neglected!
2nd issue >>cis it any way to extract the corresponding eigen values (natural frequencies) based on these vectors to make sure that these vectors are really representative to the mechanical system, where the shown methodology used in my my post proved to be unsatisfied when I tried to compare eigen values with the corresponding natural frequencies ?
Best of Regards
Mohamed A M Salem The script I sent to you was just an example to show you the methodology. If your finite element model is made of beam or shell elements including rotational degrees of freedom you have to include them in the calculation of the mode shape by paying attention to the kinematics model at the basis of the element (usually Timoshenko theory for beam elements and Reissner-Mindlin theory for shell elements).
Regarding the second question, you can find here attached the modified version of the previous script where I've added the array nat_freq collecting the eigenfrequencies that are recovered by means of the general *GET command within the *DO loop.
Best regards,
Marco Montemurro
- Badges
- Science topic