topic Re: Constrain in Modal Analysis in Nastran Forum (Read Only)

Constrain in Modal Analysis

xing.kevin — Tue, 07 Jul 2020 09:08:25 GMT

we have a apparatus be put on ground. When we did a modal analysis, we add constrain on mounting foot, constrain vertical direction. But when we did impact test, we found the modal shape don't match with FEA. The mounting foot happened moving in some modal shape. So we figure the constrain is unreasonable, maybe only limit +vertical, not include - vertical, but the software don't have this options.

Re: Constrain in Modal Analysis

John_Holtz — Tue, 07 Jul 2020 17:50:43 GMT

Hi @xing.kevin

What software are you using? I do not recognize it, so I think you have posted your message in the wrong forum.

Re: Constrain in Modal Analysis

xing.kevin — Wed, 08 Jul 2020 03:47:53 GMT

I used Nastran in cad did FEA. The first snapshot is modal test result when i use hammer do impact test to get ODS. When I did impact test, I put the apparatus on ground. When I did FEA, I added vertical direction constrain on foot in nastran in cad. The FEA result don't match with actual ODS. there is no available constrain to simulate the apparatus be put on ground.

Re: Constrain in Modal Analysis

John_Holtz — Wed, 08 Jul 2020 18:22:31 GMT

Hi,

I had never heard of ODS, so I looked it up. ODS = Operating Deflection Shape. I assume the top images that you provided are from your ODS software, which is why I was confused about what you were using for the analysis. 🙂

And to clarify, does the measured data show that the feet are lifting off of the floor? (It is hard to know what the figures show, so I think what I am interpreting as the foot on the -X side of the machine is passing through the floor is just because I don't understand anything about your images.) Or are you guessing that the difference between the measured frequency and the modal analysis is because of the feet lifting off of the floor?

About the Nastran analysis. The limitation is not the constraints. The limitation is that Normal modes (modal, natural frequency analysis) is based on the assumption that the model vibrates equally in both directions. The model vibrates as a full size wave (the feet can lift off of and pass through the floor), not some type of half wave (the feet can lift off of the floor but not pass through the floor).

If your machine is really lifting off of the floor, then you cannot use a modal analysis to simulate it. (Sorry, but I have my doubts that all N feet lift off of the floor when you hit it with a hammer. But since I have not done such a test, maybe my understanding of a hammer is different than what is used in the physical test. Any foot that does not lift off of the floor in the physical test can be fixed with a constraint.)

I think these are your options:

If you have not measured the motion of the feet (where the machine rests on the floor) in X, Y, and Z directions, you should do that. Any foot that does not move can have a constraint applied in that direction. (Obviously, feet that do move in a direction cannot be constrained in that direction.)
You need to make sure your modal analysis is accurate. Have you made the mesh smaller to confirm that the results have converged? Are the masses representing the equipment at the correct location?
If the modal analysis still does not represent the real structure because of the contact (and lack of contact) with the floor, then you can perform a nonlinear transient response analysis. Model the floor and define separation contact between the floor and the machine. Then "hit" the model with a hammer to make it vibrate, just like the real physical test.

Re: Constrain in Modal Analysis

xing.kevin — Thu, 09 Jul 2020 09:36:11 GMT

John,

You are right. The physical test result show the foot lift off from floor. When we applied Ty constraint on different position of foot (such as plane, or point, or line), we can get different result, some constraint can get similar result that foot lift off from floor. But it is difficult to exact result to match physical test.

I will try your suggestion to use nonlinear transient response analysis.