Abaqus Natural Frequencies Extraction

[jagh80]

Hello everybody!

I would like to have some help with abaqus concerning the extraction of natural frequencies. I would like to understand the outputs of Abaqus.
I analysing a gas tank made of steel plates. I'm using shell elements S4R to model the tank and I performed a linear perturbation frequency analysis to extract the natural frequencies (eigenvalues).
First, I'm obtaining in a range of 0 - 2Hz a lots (thousands) of modes, which I assume the majority are either local modes either "incorrect" solution of Abaqus. I wanted to analyse the mass participating in the modes, and for that I found the Effective Mass (EMi) for each degree of freedom on Output History of Abaqus. Though, I do not uderstand the values because either they are very small (which can be the sign of local mode) either very high (much higher than the mass of the tank). So, I'm confused. What is this Effective Mass (EMi) that Abaqus provides? What are the units? (If for example I work in kg, N, mm).

Thanks for any help.

Further information concerning the model:
height of the tank =+/- 33m
radius =+/- 29m
thickness of the wall plate = 5mm
thickness of the roof plate = 10mm

Eignesolver: Lanczos

Best regards,

Alexandre

 

[csk62]

Without knowing any specifics of the problem, obtaining frequencies at/near 0Hz could indicate the presence of rigid body modes.

Abaqus does not have "units"*. It should state the total (movable) mass of the model in the .dat file that is created. Typically, the effective modal mass is presented as a percent of total mass. So, for example, your movable mass is 10, and the effective mass in the x-direction for a certain mode is 1, then you have 10% for that mode. If you take "enough modes" into account the x,y,z sums should asymptotically approach the total movable mass.

I don't think it is possible to exceed the total movable mass in an effective modal mass calculation, so maybe you are confusing the actual mass of the model with the constrained "movable" mass of the model.

Also, something to take into consideration if you have any degree of torsional vibration is the rotational degrees of freedom. I believe that these "effective masses" are based on the MOIs.

*typically in my work, I have seen tonnes used as the base mass unit in Abaqus, mm used as the base length unit and Newton as the base force unit (giving stress results in MPa). But, I have no idea what you are using in your material data files and drawings.


All of this is from memory, but I can check some of this out if you're still having issues.

 

[jagh80]

Dear csk62 thank you for your answer.
I don't agree that the effective mass is in % because in same cases is clearly above 100% whcih is not logical. I found then that I had a mistake in copying the effective mass results and I found that the mass is for example in my case in kg, as I introduced the material density in kg/mm3. Now the results in terms of effective mass are in my view coherent. I have now to check why so many modes at frequencies close to 0Hz.

 

[csk62]

I apologize for my wording. What I meant to say was that the effective mass is displayed in [kg] or whatever unit in the dat file. However, the information is typically presented as a percent when quoting the importance of a mode, or how much of the model you have captured.

Could you attach the .dat file? I still don't think it is possible to exceed the unconstrained mass, and am interested to see this

 

[jagh80]

It doesn't exceed, after finding my mistake in the extraction of the results.

 

[bkal]

From the look of your dat file, there seems to be a problem with your model, having so many low frequency modes does not look right. Have you provided sufficient boundary conditions and connected all the parts of your model? Regarding units, working in N, kg and mm is bound to lead to wrong results in dynamic analysis. If you need to use N, kg and mm, than you should specify your density in tonne(1000kg)/mm^3 (so it is 7.85E-9 for steel) and acceleration (if required) in mm/s^2.

 

[jagh80]

Yes, I found that in order to have Hz in the frequencies, if I work N, mm then mass should be in ton.
Thank you.

 

[NunoMPG]

Here's my two cents: you should search for something called "tables of consistent units" in google it's very helpfull (i think it should be the first hit in google).