Hi,

I sorry to let you know that i don't work on isolation problems. I currently carry out my research in SSI only. 

Thank you

Hi

Operational Modal analysis, SSI, provides values on mode shape, natural frequency and loss factor. 

However, unless scaled excitation is made, e.g. response/force, the mode shapes are not scaled with respect to the system mass/stiffness. This tends to be difficultt fo large scale structures. 

This implies that one must work the problem the other way around, i.e. to build, based on first principles, a simulation model for the large structure and, within the bounds of what can  be motivated, adjust the model to what was measured. 

Some ramblings of mine on this subject can be found here: http://qringtech.com/learnmore/why-simulate-measure-correlate-automate/

and an example is found here

https://www.researchgate.net/publication/266472324_FE_Model_Correlation_Mode_Shape_Updating_using_Qualification_Test_DataA_case_study_on_the_Olympus_Satellite?_iepl%5BviewId%5D=d7XAP9T6GAjn82BlucnvxJIH&_iepl%5BprofilePublicationItemVariant%5D=default&_iepl%5Bcontexts%5D%5B0%5D=prfpi&_iepl%5BtargetEntityId%5D=PB%3A266472324&_iepl%5BinteractionType%5D=publicationPreviewImage

When tuning a model to test data, it usually is prudent to keep the following in mind

• The quote by G.P. Box - all models are wrong, but some are useful - which applies both the the test and the simulation models. 
• Identifying test modes from test data is an inverse problem, i.e. a problem of the type where the answer is known, say 4, and your task is to identify the question. This implies that it is easy to get any set of test modes but difficult to get a fit and proper set of modes. I usually make 40-50 modal analyses using various assumptions, throw out part of the data and so one before accepting a data set. Also, as you can see from the project example, it is not a given that the test object always behaves the same way, i.e. has linear and time invariant properties. In fact, most structures do not.  
• One must match test geometry with simulation model geometry. 
• When tweaking a model, things like Young's modulus is know to good accuracy. Other things like interface stiffness, differences between assumed and actual geometry tend to matter more but are all the harder to describe. 

Hope this helps

Claes

Article FE Model Correlation & Mode Shape Updating using Qualificati...

Please check reference below.. hope it will be helpful for you… Regards…

https://nptel.ac.in/courses/105108069/mod09/lec09.pdf

Good question