Learning Forum

[hakankocak0]

I am trying to prepare a modal analysis for a complicated system. I only modeled the shell of my system, and I am going to use point mass to define the mass and inertia of my subsystems. However, I was not sure about the accuracy of this method; hence, I modeled a simple cylinder with a diameter of 50mm and a length of 300mm, and calculated its inertia and mass. Then I used two methods to find natural frequencies.
1. I assigned original material and solved it.
2. I assigned a material with very small density(1 kg/m^3), and I use point mass to define mass and inertia. I used "deformable" for point mass. (I am planning to use this method for my complicated system)

I expected to have similar natural frequencies. However first one resulted around 2300 Hz, and the second one resulted around 200000 Hz.
Then I divided the cylinder to two, and define two point masses. After I divided the cylinder two, the result became 2400 Hz.
Then I divided the cylinder to four, and define four point masses. After this try, I obtain natural frequencies which are similar to the method that I assigned the original material.

In conclusion, I understood that increasing the number of point masses increases the accuracy. However, even for a 300 mm cylinder, 1 point mass results in a non-reasonable value. How can I know that I use enough point masses? Or, is there something that I miss? Thanks for your help, in advance.

[dlooman]

The deformable behavior distributes the mass without stiffening the cylinder, but it can lead to ill-conditioning.   Is the cylinder modeled with solid elements or beams?  How did you scope the point mass to the cylinder?  

[hakankocak0]

I modeled the cylinder as a single solid body in SpaceClaim and then partitioned it into two and four sections. It is solid, not hollow. I scoped the point mass to the outer curved surface of the cylinder.

I observed that the scoping area of the point mass has a significant influence on the results. Decreasing the area of the selected faces reduces the natural frequency. However, I am unsure about the correct approach. Which faces should I select? For instance, if I have an electronics box and I only model the housing, should I apply the mass and inertia of the electronics to the fastening/mounting interfaces, or should I scope the point mass to all internal faces of the box?