Discussion Forum

Hi

recheck your boundary conditions, are you sure you have a single solution. And is your structure fixed somewhere ?

--
Good luck
Ivar

Hi , thanks for ansering my question .

My structure is fixed at the base .

When i am running a static eigenvalue analysis i have only to enter the boundary conditions and the properties of the material , right?

Did you put the material densities? Zero values could cause that error.

Yes of course ,
i have set all the material properties ,
density , poisson ratio and young modulus .

the steps i ve done are :
1. Start a Structure Mechanics 3D Static Analysis
2. Import my structure
3. Set all the material properties(E , v , p)
4. Create identity pairs .
5. Set the constraints ( base of my structure is fixed and also the boundaries of the identity pairs are fixed)
6.Mesh
7.Choose from the solver parameters eigenvalue solver .


Then when i sovle the problem appears "mass matrix is zero" .

It could be the way you select the eigenvalue solver (from the Solver parameters as you say). COMSOL may still be considering this a Static Analysis as you defined in Step 1. It’s best if you select the Eigenfrequency analysis directly from the Preset Studies (instead of Step 1). I hope that helps.

Nagi Elabbasi
Veryst Engineering

Hi
Or is it that you use nu = 0.5, if so try 0.49 or smaller

--
Good luck
Ivar

Thanks for reply ....

Yes that's true ,when i select eigenfreqiency analysis at the first step it runs fine without errors , but i am not sure if the results i take from this analysis are the same for static-eigen value analysis....

I need the results from eigenvalue analysis to run a dynamic analysis of a seismic load on my structure .

thank you but what is nu?

Hi

for me nu, here, is the Poisson coefficient of the structural material, typically 0.3 in anycase 0<= nu < 0.5 (but COMSOL fails if you try nu=0.5 for obvious physical reasons.

As I had few clues to guess what it could be I send out some of the effects I have encountered previously on the Forum, not everyone reads carefully the doc before modelling ;)

--
Good luck
Ivar

Hello Every one

Based on my background in Mechanical Engineering, the value of nu needs always to be less than 0.5, as Ivar stipulated before. Because in the Structural mechanics the expression "1-2*nu" shows up in the equations; and it needs to be positive, always. Whence, nu=<0.5 should be satisfied otherwise the basic equations fail!

Good Luck

Good points Vahid, Ivar,

For what it’s worth I have some comments on the perfectly incompressible case of Poisson’s ratio exactly equal to 0.5. The solid mechanics equations can theoretically still be solved in that case. A mixed formulation where there are pressure and displacement degrees of freedom is required, and the equations have to be recast in a way that does not have 1-2*nu in the denominator. I do not expect to see that in any commercial FEA code because it’s not useful/practical. I believe that is due to three reasons: (i) it’s not worth the hassle since nu=0.4999 works well with a mixed formulation, (ii) it involves zeros on the diagonals of the pressure degrees of freedom which solvers don’t like, and (iii) no real material is perfectly incompressible.

Nagi Elabbasi
Veryst Engineering

Hi,

It seems to like this thread (although old) is going slightly off topic. The mass matrix will not be affected by the elastic data.

It is not necessary to have any constraints at all in an eigenfrequency analysis, so we can also rule out the boundary conditions..

Having a zero mass matrix is rather odd. The only material property that enters the mass is the mass density. If it is non-zero and correctly assigned to the domains, this should not happen.

Regards,
Henrik

Hello,

I have modeled a simple cylindrical tube using pressure acoustic, frequency domain study. I have tried to extract the state space matrices using MATLAB Livelink. The problem is the mass matrix turns to a Null matrix. I have used sound soft boundary condition for the outlet, Pressure boundary for the inlet, and the other boundaries are sound hard boundaries.

I will be really thankful if someone gives me some hints regarding solving my problem.

Kind regards,

Shahin Mohammadnejad