Modal Participation Factor Meaning in Nastran

[Francesco Bro]

Hi all!

We have been running modal dynamic simulations for an aerospace structural system, and I run into a basic, even tough tricky question: what is the physical meaning of modal participation factor (MPF)? What is the difference in terms of physical meaning with respect to modal effective mass (MEM)?

I understand how it is computed from theory.

MPF = [PHI]^T [M] {r} / [PHI]^T [M] [PHI]

MEM = ([PHI]^T [M] {r})^2 / [PHI]^T [M] [PHI]

I have always heard in the industry about the MEM (or MEMF), how important is to extract modes to up to at least the 85%-90% of the cumulative MEMF, and its meaning: gives us a sense in which "global mass" is moving in each direction in T1T2T3/R1R2R3.

But what about the MPF? Is it used in the industry? What information does it give us from the physical meaning?

On the other hand, I found 2 separate case control commands to request it in Nastran:

$ CASE CONTROL SECTION
MEFFMASS(...,PARTFAC,...) = YES
$
PARTN=n

What is the difference of each case control command? When should I use each one?

Thank you!

 

[nicomedian]

I would highly recommend to read Tom Irvine's notes on this issue to gain insight and learn how each and everything calculated.

 

[Francesco Bro]

Hi nicomedian. I did, but it says nothing about the physical meaning, physical difference between MEM vs. MPF, industry standards, requirements, etc.

 

[nicomedian]

Physically, what I learnt from Tom Irvine's notes is the following: These numbers give you a measure, how much you can excite the systems CoM by applying base excitation. In other words, it gives you a measure of which modes might play important role for your analysis/test.

That's why there is no mass participation for free-free systems, because base excitation is not possible. If you request the output, you will see that all the mass value will be collected in rigid modes, which means you can excite complete mass of the system by the rigid mode. Remaining elastic modes will be practically zero for mass participation.

You may find some standards in the industry. I can give one example from ECSS‐E‐ST‐32‐11C page 49.

I hope this may help to your quest.

 

[dmapguru]

Modal Participation Factors (MPF) have several definitions; the one defined in MSC Nastran follows the civil engineering definition you gave in your post. This is projecting the eigenvectors onto the rigid body vectors {r} and comparing it with the eigenvectors projected onto themselves (which is 1.0 if mass normalized eigenvectors are requested). So physically this is giving you a measure of what proportion of the mass is moving, compared with the total mass that moves in the rigid body vector shapes, for each mode. Local modes would show low MPF and global modes would be higher.
Be careful though, as already mentioned, MPFs calculated using the civil engineering definition *must* have a statically determinate boundary condition (STBC) defined for the structure (of which base motion is a particular case where all DOF at the same GRID point are restrained). If this STBC is not respected, the (civil engineering definition of) MPF is meaningless.
As you stated in your formulae, the MEM is just the square of the MPF – some people prefer to report one and not the other, so MSC Nastran may output both for convenience.
In my opinion, MPF is a poor indicator of sufficient modal richness for dynamic response.
DG

 

[Francesco Bro]

Thank you dmapguru for your response!

Could you please help me by clarifying they difference of the MPF request in the Case Control?

$ CASE CONTROL SECTION
MEFFMASS(...,PARTFAC,...) = YES
$
$ vs.
$
PARTN=n

 

[dmapguru]

Do you mean what's the difference between the MEFFMASS command (using the PARTFAC option) and the PARTN case control command?

 

[Francesco Bro]

Yes.

 

[dmapguru]

MEFFMASS is computing modal participation factors as a step to yielding the modal effective mass (which, because of the mass normalization of the eigenvectors are the square of the MPFs). So here, the participation is the contribution each mode has to the total possible movement of the rigid body mass in each of the 6 basic directions in which the structure is able to move as a rigid body.

Now, back in the bad old days, when you ran a SOL 111 modal frequency response in the presence of a fluid cavity, MSC Nastran would compute modal participation factors. Most unfortunately, this is the same name, but it is not at all the same thing. In the SOL 111 case, the modal participations were the contribution each mode was making to the response of the structural GRID point response, like an acceleration (if you were considering structural modal participation factors) or the modal participations were the contribution each mode was making to the response of the fluid GRID point response, like a pressure (if you were considering acoustic fluid modal participation factors). I just don’t remember what the scope of the modal participation factors was, but if it was all GRID points and all modes, that would be a lot of output. The PARTN= command could be used to point to a SET definition which listed the GRID points where you wanted the participation factor output, a way of reducing the data to a sort of panel. This methodology was obsoleted out in about 2004 when the FLSPOUT command came along where you could decide, on a very elaborate interface, which participation factors you wanted and where in the model. Since then, FLSPOUT has been obsoleted out by the PF* set of commands which perform the same computations orders of magnitude faster (with a slightly less elaborate interface, as they separated out the different types of participation factors onto separate commands).

Today, PARTN= has no link to participation factors because it is done another way, and it only gets used by the DMAP writer (see MATMOD/17) who needs a way of defining a user defined set of GRID points to make a partition vector to partition matrix data. Even then, there are better (easier) ways of doing this either through the SET1/SET3 interface or via the PANEL definition interface, so PARTN hardly ever gets used nowadays. Something is telling me that PARTN= also got used for another thing, but I can’t remember what; it wasn’t participation factors anyway.
DG

 

[stefanix1972]

Dear all,
I just came across in this thread because I'm interested in calculating the modal effective mass of a portion of my structure.
In Altair Optistruct MEFFMASS command is "cloned" and improved:
MEFFMASS(COMP,PROP,SET=1001,SUMMARY,PARTFAC,HG,OPTI)=YES
it is possible to define a set of elements to get their MEFFMASS.
As far as you know, is there the same possibility in Nastran?
From this thread it is not clear to me.

Kind regards,
stefanix1972.