Sub-Component Dynamic Analysis

[Spirit]

Hello forum,


our customer, a manufacturer of two-stroke engines, has asked us to perform a dynamic analysis af an electro-mechanical device mounted on their engines.
For this purpose, they have placed some accelerometers on the mounting locations of the equipment and recorded the acceleration time histories during engine use.
We are going to do a FE of the component, but how are we supposed to use the acceleration data? Should we use directly the time histories (which are huge data files) and perform a transient analysis or approach the problem in some other way?

Thanks for your assistance.


Regards

'Ability is 10% inspiration and 90% perspiration.'

 

[CESSNA1]

SPIRIT: On way is to use the time history data as an input into the FEA model. That can be cumbersomne if they want a long test time. For what are they looking? Is this a transient analysis or a resonance search? They could use the time history data as input into a shaker rather than the FEA model. Will the vibration data be in one, two, or three directions?

Regards
Dave

 

[Spirit]

CESSNA1,

vibration data are in three directions.
The objective of the analysis is not to exceed prescribed accelerations and displacement values in the attached device that we should simulate. I was wondering if there is some way to move from time domain to frequency domain for the input acceleration that we should give to our model. I was thinking of Fourier transform of acceleration data, but what do you use it for once you have it? Frequency response analysis ...? Transient response ...?

Thanks, regards.




'Ability is 10% inspiration and 90% perspiration.'

 

[cbrn]

Hi,
Spirit, yes, as you say it's far better to find the acceleration spectrum and shift to frequency domain. Then, you will perform a "response spectrum analysis" with your FEA. BUT...
1- first of all, your FEA must support this
2- then, your model must contain NO non-linearity

If the second condition is not respected, you will have to revert to a full transient analysis (in fact, as far as I know nowadays no implicit solver is able to do a spectrum analysis with non-linearities).

Regards

 

[Spirit]

cbrn,

the software is NASTRAN, I do not have previous experience in response spectrum analysis though...
As far as you know, is it suitable?

Thanks, regards.




'Ability is 10% inspiration and 90% perspiration.'

 

[cbrn]

Hi,
I really know nothing on NASTRAN, I'm sorry...

Regards

 

[GregLocock]

As far as I know it is suitable but you'll need some sort of fatigue postprocessing.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.