Modes of Vibration: Excitation 1

[ansysnewbee]

Hello All,

Iam trying to get some modal results (nat freqs and damping constants) for a frame type structure.

Iam using a shaker excitation (periodic) to excite my structure in the 0-500Hz freq range. The problem Iam suffering is that, when I excite the structure in say Z direction, I miss the axial modes of the structure (X direction modes), and when I excite the structure in its X direction, I miss the global modes, but succesfully capture the axial modes of the structure..This is could be partialy because of the higher stiffness of the structure (am I right?)

I would like to excite my structure at a point where I can extract all the modes in one go. Using two shakers is not an option(dont have two shakers!).

Anyother way around it?

Any advice would be very much appreciated.

Emily.

 

[GregLocock]

You have several alternatives, these numbered in my order of preference

2) hire another shaker and do a MIMO modal

1) do more than one survey (this is good practice, it allows you to check reciprocity), shaking in different directions at different locations.

3) Mount the shaker so that it is driving at 45 degrees to each plane

It is not due to the high stiffness, it is because the mode shapes have little coupling into the other directions.

The disadvantage of method 3 is that the results are no longer easily manipulable mathematically, unless you use the same coordinate system for your response points (or I suppose you could still use the obvious coordinate system for yuor accelerometers and then synthesise responses in the shaker coordinates)





Cheers

Greg Locock

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

 

[ansysnewbee]

HI Greg,

Thank you for your response..

this raises a few questions (sorry to bother you again!)

Is there any mathematical explanation for this? Is there a name for such a behaviour? or any better explanation? or is it just a *common thing* in experimental modal analysis..I tried doing some searches, but could not find anything useful...

Meanwhile, iam going to look at some prices for hiring an additional shaker and do a MIMO.

Thank you.

-Em

 

[GregLocock]

"Is there any mathematical explanation for this? Is there a name for such a behaviour? or any better explanation? or is it just a *common thing* in experimental modal analysis..I tried doing some searches, but could not find anything useful..."

You'd also see it in synthesised dynamic responses from an FEA of the same system, so it is not exclusive to EMA.

Choice of excitation point (and direction) is a very significant issue for many real applications, I've never seen any very helpful remarks - so I guess you can add this thread to that list!.

Ideally each excitation point would excite all the modes in the structure in the frequency range of interest, at roughly the same amplitude.

Worst case I can rememebr is doing a mini survey, looking at responses from 5 different excitation points and directions at say 20 response points (also checking reciprocity as I went).

Then I picked the two best and used those for full surveys.

A typical problem is when a system contains an isolated sub system eg a full vehicle has the engine mounted in it - we are quite interested in the coupling of engine flexural modes and those of the floorpan, so in the bad old days we'd try and do a modal of both at once. Trouble is, with 30 dB of attenuation across the elastomeric mounts it was not really possible.





Cheers

Greg Locock

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

 

[MikeyP]

If your exciter position is at a point on the structure which doesn't move much for a particular mode shape, then you cannot put much energy into that mode shape. In the extreme case, if the exciter is at a nodal point or on a nodal line (that's a mode shape "node", not an FE "node") then you cannot exite the mode at all (unless energy can be coupled between the modes by for example non-proportional damping or non-linearity).

Here's a simple example: the ideal beam which is simply supported at both ends. If you put your shaker at the mid point of the beam, then you cannot excite modes 2, 4, 6, 8 etc because the shaker is at a nodal point for all symmetric modes.

A similar argument applies to your problem. If there is no mechanism to transfer vibrations in the x direction into vibrations in the y direction then a structure excited in the x direction will not respond in the y direction. In reality, there will probably be SOME response in the y direction, but it will be small and prone to corruption by noise.

There are a number of ways to select exciter positions. One is to do a "quick and dirty" modal survey using a hammer. That way you can quickly try various exciter positions without having to set up the shaker each time. A second approach is to use an FE model. If you are reasonably confident that your FE model is OK then you can try different exciter positions on the model.

I recently did a modal survey on a 1/50 scale model of a large civil aircraft. I used SIX exciter positions, some x, some y and some z. When I came to do the modal analysis I got some pretty crummy results. The reason was this: The exciters in the x direction were contributing very little to the modes which responded predominantly in the y direction. The FRFs between an x exciter and a y response were very noisy with poor coherence. So during the modal analysis when I was attempting to curve-fit a mode with a dominant y response, I only included the FRFs measured from y excitation positions. Voila! pretty looking flappy mode shapes.

M

--
Dr Michael F Platten

 

[GregLocock]

What do you think of the barbarous practice of mounting the shaker out of the plane of symmetry?

Cheers

Greg Locock

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

 

[MikeyP]

It is not something I have tried so I don't have the practical experience. The problem with that is how to accurately and precisely do the coordinate transform to get the "true" force in each plane. I would be much happier with it if there was such a thing as a triaxial force gauge that I could put on the end of my stinger. Maybe there is, but judging by the lack of google hits for "triaxial force transducer", their use is not widespread. Anyway, why bother? It is really so much effort and cost to stick another shaker on?

I know that "compound angle" attachement is common in the road vehicle world, but I've never seen it on an aircraft test. Many small shakers instead of few big ones means the same energy input with less force drop-out and less chance of local non-linear deformations at the attachment points. Also the problems of noise I described in my earlier post would be nothing like as severe if I had the latest 24-bit ADCs and the latest curve fitting software (The new multi-reference z-domain methods are much more forgiving of noise).

M

--
Dr Michael F Platten

 

[MikeyP]

Greg,

Having actually properly read your last post (!), I think I may have answered a different question to the one you asked. What do you mean by "out of the plane of symmetry"? In a symmetric structure, I would ALWAYS put the excitation point away from a point/line/plane of symmetry unless I was deliberately trying to suppress particular modes.

M

--
Dr Michael F Platten

 

[GregLocock]

No, I meant compound angle excitation. It's so long since I've worked in the modal lab I'm forgetting the technical terms! (Bliss)





Cheers

Greg Locock

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

 

[ansysnewbee]

Oh..This post is getting very inetresting..and Iam learning a lot of new stuff!!

Thank you very much Greg and Michael F Platten..It has answered a lot of questions.

I would be interested to know how to process the data if the shaker is at a compound angle.

Awaiting more discussion...


-Em

 

[MikeyP]

Like I say, I have never actually done it, but if you just want natural frequencies and modal critical damping ratios then I see no reason why the data processing should be any different.

However, if you want a full modal model (ie including mode shapes and scaled modal masses) then you need to have a "driving point FRF" measurement. This is where the applied force and measured response are determined at the same point AND IN THE SAME DIRECTION. So if you haven't measured a driving point FRF (ie your response accelerometer and your force trancducer do not line up) then you need to determine it, either by decomposing the force into the coordinate system of the response transducer or transforming the response measurement into a coordinate system which is aligned with the force transducer. Some experimental modal analysis software will take care of all this for you.

M

--
Dr Michael F Platten