DMAP Nastran Centrifugal Force

[vanji1]

Hello

I have a finite element model of a bladed-disk and I put a centrifugal force in order to take into account the modeling the rotating speed and the stiffening of the structure (In order to calculate the eigenfrequencies).
I have no problem with the Nastran calculation but I would like to extract the generated centrifugal force and/or the static displacement induced by the centrifugal force.
I am not an expert with DMAP and I would like to know

-- how to know which subprogram is concerned (If I understand well SEMODES) ?
-- what is the name of the variable to be extracted ?

Thank you in advance,

Evangeline

 

[dmapguru]

You don’t need DMAP to extract the quantities you seek (although I could tell how to do this, it would be a little convoluted). You can use the OLOAD=ALL case control command to output the computed forces/moments from the static load and the DISPLACEMENT=ALL command to get the displacements. Specifically:

SOL 103
CEND
SUBCASE 1
OLOAD=ALL
DISP=ALL
LOAD = 11
SUBCASE 2
STATSUB = 1
METHOD = 12
BEGIN BULK
RFORCE,11,…
EIGRL,12,…

For information, you would not need to run SOL 103 to get these quantities; you could just run SOL 101 with the first SUBCASE:

SOL 101
CEND
SUBCASE 1
OLOAD=ALL
DISP=ALL
LOAD = 11
BEGIN BULK
RFORCE,11,…

This would avoid having to wait until the eigenvalue computation completes to get the results.

Both OLOAD and DISPLACEMENT outputs are written to the post processing units (op2/hdf5) or can even be punched to the punch file. This would allow you to load them into a post processor and plot vector or contour plots.

On the subject of stress stiffening in rotating structures, do you realize that the procedure above will extract the eigenvalues of the stress stiffened structure as if were not rotating? This is one of the crazy things you can do with simulation. The static load comes from the centripetal acceleration due to RFORCE, but the structure is not actually rotating. When you compute the natural frequencies in the presence of STATSUB, the static load is used to compute the stiffening effect of the load which appears as an additional stiffness (the differential stiffness). If the structure were rotating, there would be additional effects (which are loosely referred to as gyroscopic effects) which cause some of the natural frequencies to split into 2 frequencies, one which is greater than the stationary natural frequency and one which is smaller (these are called the forward and backward whirl modes). In a sturdy rotating structure, or one that rotates at a slow rate, the gyroscopic effects are minimal as they are a function of out-of-plane displacements of the rotating components. In flexible rotating structures, or structures that rotate at higher speeds, the gyroscopic effects can be significant and can make the natural frequencies change a lot.

DG

 

[vanji1]

Dear DG,

Thank you very much for all these informations. Thanks to you I understand that the structure does not rotate so the stiffening effects only comes from the linear static calculation with the centripetal force due to RFORCE, which I did not understand.
Thank you also for the trick that could avoid the calculations of the eigenfrequencies and allowing to this information to be get.

From the .f06 file, I am able to read the output displacements without the Dmap and get also this RFORCE by doing the
static calculation, having extracted KE + KDIFF. So I have to understand why the Nastran RFORCE formulation and my direct fe calculation do not yield the same results…

I also have another question. How is generated such differential stiffness KDIFF ? Are the stresses first computed from the static displacements first ? Where can I find this information on the Nastran manual ? and on the DMAP PROGRAMS ?

I thank you in advance,

Best regards,

Evangeline

 

[dmapguru]

You said you have read the displacements and the RFORCE loading (presumably from the OLOAD=ALL output), but you also said “having extracted KE + KDIFF”. How did you extract this stiffness? Be careful, in static analysis (SOL 101), differential stiffness is not computed. To ensure the differential stiffness is computed (and thereby extracted), you need SOL 103 with a STATSUB definition as I illustrated previously (other solution sequences will also generate the differential stiffness, but you already have a SOL 103 job prepared) and you need to extract the stiffness at the right point in the series of operations.
Note that the differential stiffness may be generated with or out without follower force effects. For RFORCE loading, this is controlled by the RFORCE entry via the METHOD field (field 9) which can be defined with a value of 1 or 2. If defined with a value of 1, follower force stiffness will not be generated, only differential stiffness. Differential stiffness can be appreciated by thinking of a guitar string being tightened where the string length (distance between supports) remains constant. With increased tension (load), the string vibrates at a higher and higher frequency. If METHOD is defined with a value of 2, the follower force stiffness is added to the differential stiffness. Most people think of follower force effects as nonlinear as the name suggests that the forces “follow” the deformation of the structure. This is also the case in RFORCE loading, except the direction of the forces does not change, but the elastic deformation of the structure means the mass moves outwards under the action of the centrifugal loading. Think of a uniform disc spinning at a constant speed. The centrifugal force can be thought of as point loads acting on little pieces of mass of the disc all of which act to expand the disc, that is each little piece of mass of the disc increases its radius under the action of the centrifugal force. This increase in radius actually reduces the apparent stiffness of the disc and in some literature is known as spin softening, but what is important is the combined effect of the differential stiffness and the follower force stiffness, and the two result in an apparent increase in stiffness but not as much as the increase in stiffness from the differential stiffness alone.
Differential stiffness and follower force stiffness (if requested) require a displacement field (which is computed from the static analysis) and an element formulation. The theory of differential stiffness is somewhat rudimentary and based in beam theory. You can find the theory in some text books, but the best description I have seen is in the MSC Nastran Theoretical Manual, chapter 7. Search the IOT for the theoretical manual and you will probably find the NASA document.
If you have only the task of extracting differential stiffness, learning DMAP will be a heavy deal. If you tell me what you want to do exactly, I can help you with this. If you want to learn generally, then there are a couple of resources for DMAP programming. The DMAP Programmer’s Guide helps you with data structures and modules (it’s free as well), but does not provide a tutorial for learning the language. MSC provide paid training courses on the subject and there is a book about it by Mark Robinson. Again search IOT for Robinson DMAP and you should find it. There are several examples of DMAP alters delivered with MSC Nastran, in the sssalters directory.