PrasannaK liked a post in a topic by Farouk Maaboudallah in Model analysis
First, you are running Real Eigenvalues Analysis using Lanczos solver (EIGRL).
V1 and V2 mean the lower and upper bound of your frequency domain.
ND means the number of roots (number of eigenvalues) of the following equation :
(-w^2*M - K)*phi = 0
with w refers to the real eigenvalue and phi means the eigenvectors.
oguzhan liked a post in a topic by Farouk Maaboudallah in Contact between 2d and 3d
Hi mate !
You can use for example TIE constrain to link the master surface with the slave one.
Normally, in contact problem we refine the mesh of the slave surface. But I think in your case, linear static analysis, there will be not so much difference. You can use the automatic mesher and mesh the parts using Tetra approximation.
It's okay if you use Tetra finite element ... But in Nonlinear static analysis, we recommend to refine contact interfaces (slave surface) with S2S discretization to ensure the convergence of the contact pressure.
You welcome mate
PrasannaK liked a post in a topic by Farouk Maaboudallah in Contact law implementation
Thank you for your answer
Yeah I now exponential, cubic laws .... The idea is for each NSET (Contact interface) we compute our own law ...
For example, we assume that we want to perform contact simulation between 2 bodies (see the model below). The idea is to divide slave surface into 6 sub-surfaces and for each sub-surface we want to use a specific contact law (house development).
Rahul R liked a post in a topic by Farouk Maaboudallah in How is VonMises strain output in hyperview calculated by Optistruct?
Joel Rodarte liked a post in a topic by Farouk Maaboudallah in How is VonMises strain output in hyperview calculated by Optistruct?
I hope it will be clear for you
Farouk Maaboudallah liked a post in a topic by Rex in Fatigue simulation in frequency domain
1) ASD is Acceleration spectrum density, it always with the unit (G^2/Hz) in X direction and Hz in X direction, and it tells how much the acceleration will be in 1sigma level.
2) If you want to know why I want to run the random fatigue in frequency domain, you can search on google, they explain it better than me. For me, I would say do the random fatigue in time history is a very time-consuming work, and usually if you want to know the random input in a better way, you need enough acceleration data in time domain. Meanwhile, if you see the acceleration input on time history, sometime it does not help you too much, but in the frequency you could find which frequency is more important.... oh, by the way,by using Miner rules, we can even get the total damage considering different acceleration inputs(like the acceleration on the air intake manifold can be caused by engine and road together, it's difficult for us to consider the fatigue in time history...)
Please learn more on this topic by using google or here, all these answers are based on my understanding, if I am wrong, just let me know!
saiful85 liked a post in a topic by Farouk Maaboudallah in Transient Non Linear Analysis (Contact)
There a good book provided by Altair. Inc. You can find the basic of what you're doing. You can find as well the model described in details ...
Geraldo Osorio liked a post in a topic by Farouk Maaboudallah in PARAM, FFRS, FASTFR
FASTFR : This option is used to activate "The faster modal solution method" for modal frequency response analysis !
FFRS : The second means FAST Frequency Response Solver, used if you want to activate the external FASTFRS solver. This solver works well for a large modal frequency response. (You have to install it in advance).
FFRSNCPU : used with FFRS. It's useful if you want to define the number of CPUs to be used by the external solver "FAST Frequency Response Solver".
FFRSLFREQ : It's a threshold (cut-off frequency) used to partition your system into low freq and high freq.