Beam Bending 1

[creighbm]

It appears that NASTRAN can not compute bending due to an axial load...is this true? For instance, I calculated the deflection of a beam due to a lateral and axial load. When I modeled this in NASTRAN, it seemed to neglect the axial load and only calculate the deflection due to the lateral load. Is there a way around this (I am using CBEAMs). Thanks.

 

[GBor]

...bending due to an axial load[\quote] Isn't that '0', or am I missing something:

Sigma bending = Mc/I...no axial load involved, right?

 

[creighbm]

Not really. If you have both axial and lateral loading, then the deflections/moments will be greater than if you have lateral loading only (assuming the axial load is in compression...according to Roark, Table 10, case 2a for example).

 

[GBor]

Ah...compression -- tension...my bad. Wasn't thinking...had a different image in my mind...need to broaden it

 

[rb1957]

you're talking about beam columns ?

the beam column moment is due to deflection of the beam under lateral loads ... so you'd need to run non-linear nastran to see this effect

 

[EdR]

Typically beam elements do not include deflections due to axial load in their basic definition.....You must use large deflection option to get this type of secondary load included....

Ed.R.

 

[creighbm]

Thanks Ed.R.

Dumb question but is this a 'flag' in NASTRAN where I can enable this option?

 

[rb1957]

there's a flag for shear deflection, but you can't flag this effect ... it's a different SOL.

alternatively you can run linear SOL101, then deform your model according to the deflections and run again (and again)

 

[creighbm]

Well that being the case, would it be correct to determine the deflection of the beam due to lateral loading first, then determine the additional bending loads to the axial load acting offset from the undeformed beam centerline? According to Roark, deflections due to axial and lateral loads cannot be superimposed so I am guessing my method is flawed.

 

[rb1957]

no, the axial loads magnify the deflections due to lateral loads.

what you can do is start with the lateral loads, calc the deflections. these deflections, together with the axial loads, increase the moment in the beam, which increases the deflections ...

so if you model the deflected beam (deflections due to lateral loads) and apply the axial loads to this you'll get larger deflections. non-linear FEM solves this properly, i'm suggesting a way around using NL (since i'm guessing you can't). the difference between these two runs gives you the 1st (largest) amount of deflection due to axial loads (should be less than the deflections due to lateral loads). i'd then model this delfected beam (displace the beam = to the difference between these two runs) ... this is the 2nd order affect of the axial loads, and should be less than the 1st ... etc

clear as mud ?

 

[creighbm]

The mud is actually turning into murkey water now! We do have the non-linear static and transient solvers in NASTRAN, so if I understand you correctly this non-linear solver will take into account this extra bending due to the axial load in the CBEAMs?

 

[rb1957]

should do, you want geometry non-linear (maybe large displacements) so that it recalculates things based on the deflected shape.

good luck
(look up beam columns in Bruhn, or Niu)

 

[creighbm]

Got it, getting the results I am expecting. Thanks a lot rb1957.

 

[creighbm]

Ok...one more question. Now that I have the static nonlinear case to work, I am trying to perform a nonlinear time history simulation. I assumed this could be done with the nonlinear dynamic response solver but I don't get any ouput. Is there a trick to get this to work?

 

[rb1957]

if you're not after many steps, you could do several NL runs with the load you expect at the time ... or you could ask your sometimes helpfull support people

 

[creighbm]

The model has about 500 time step so I went for plan b. Apparently the nonlinear dynamic solver doesn't know how to handle CELAS2 elements (but the nonlinear static solver does...) Back to the drawing (or modeling) board!

 

[rb1957]

yeah, funny how that happens ... i've got a problem where i'd like to combine CELAS or gaps/contact with column buckling ... apparently you can one or the other but not both.

again, maybe a few NL runs of differeent times will help you understnad what's going on.

maybe there's another solver out there that'll do both ?

 

[abusementpark]

creighbm, what are you trying to analyze? Usually, the governing code for the material of the beam has built-in factors in the code to allow you to check design based on results from a linear analysis. These are called moment magnification or moment amplification factors, depending on the code, and they are supposed to be a practical approximation (albeit conservative) of the additional stress induced in the beam due to second-order effects.

 

[Struct71]

No linear code will be able to to this.
Loads on the deformed shape behave significantly different from the same loads on the original shape.
You will need to have a large-deflection switch set.
This will re-apply the loads on the deformed shape (and repeat until answer is sufficient).

 

[creighbm]

abusementpark,

I am basically modeling an accelerating beam under a 1g lateral load.

eelco,

Yep, I quickly found that out and have been able to get the NASTRAN results to match theory but performing a nonlinear analysis w/ large deflections enabled.