DNV - Yield Check Allowable Values - How to include plate bending stress?

[EngineerMickeyMouse]

Greetings Folks,

Important question. I am using general FEA software, now I have started work with DNV, and I am using DNV-OS-C201 with (page 41) yield check usin von Mises stress for plate finite elements in my model, but DNV states allowable stress for membrane stresses only, there is guide note, that one MAY use DNV-RP-C201 to account also for bending stress but it does not give transparent allowable check value.

To what compare my bending stresses, better, total stress (bending+membrane)?

 

[Crannog]

Morning,
I might be misunderstanding your problem but I would have thought the easiest answer would be to compare the FEA equivalent stress to the equivalent stress allowable?

I believe the check in RP-C201 is a localised yielding check for a plates structure. It's relevance will likely depend on the geometry and loading.

I would consider σ.x = σ.bending + σ.direct but I don't recall it that is what OS-C201 expects.

D

 

[EngineerMickeyMouse]

Crannog,

Thanks for participation in discussion.
Well, what you have wrote "to compare the FEA equivalent stress to the equivalent stress allowable" it is under investigation by my query.
DNV define value of allowable stress only to be checked against von Mises build with membrane stresses.
There is no yield check defined to include also bending stresses in plate finite elements, hence my question still valid.

 

[Crannog]

Typically the FEA equivalent stress would be conservative as it will include σ.x σ.y & σ.z, inclusive of bending. So long as you inside the allowable your design doesn't have any issues unless you are trying to squeeze some additional capacity out of it?

Depending on the geometry your only concern would be if you had a localised bending issue with a lower allowable state.

D

 

[Crannog]

I think the intent of this notation is that σ.x & σ.y contain the global bending induced stresses (i.e. hull bending) but not those from in-plate-thickness effects (such as lateral loading from hydrostatic pressure), that would be checked separately (Section 5). How you approach them would likely depend on what your situation looked like, do you have local lateral effects (particularly pressure)?

Globally considering the σ.j value as σ.e from the FEA would still be conservative (unless that poses a problem).

D

[edited to make more sense]

 

[EngineerMickeyMouse]

DNV excluded bending stress to be checked within buckling analysis.

But it is not confirmed that simply von Mises with all X Y Z components check against allowable will be safe assumption.
It might turned out that allowable stress which would include membrane and bending effects would be smaller value than just membrane requirement!
So you would compare grater stress value at the element against smaller then current allowable value... It is certainly not safe.

 

[Crannog]

I don't see many cases where global σ.membrane + σ.bending are going to be lower than membrane on it's own? And those that are (eg axial tension & compressive bending and visa versa) would be clear to identify from your loadcase?

So long as the sign for your stresses is the same then total equivalent stress (σ.membrane + σ.bending) will always exceed σ.membrane only equivalent stress? If you were very concerned with the a mis-matched loading, then you could always manually extract the membrane stress and the bending stress for the area of interest and ensure the combination of the two was additive.

But assistive loading (σ.tensile - σ.bending etc) is a genuine effect and it could be viewed as excessively conservative not to consider it, particularly if your design required capacity (and if the design is far from capacity you can afford to be overly conservative if it will save on analysis time)?
---
As I say, I believe the DNV guidance note is referring to ignoring through-plate bending rather than globally induced bending stresses, they're checked individually.


D

 

[Crannog]

If it's a help RP-C202 defines membrane stress (σ.x) as:

σ.x.Sd = σ.a.Sd + σ.m.Sd

Where:
σ.a.Sd = design membrane stress in the longitudinal direction due to uniform axial force
σ.m.Sd = design membrane stress in the longitudinal direction due to global bending




D

 

[EngineerMickeyMouse]

Not exactly, RP-C202 considers cylinder like shells as a revolution objects. It is not coherent with plate finite elements we are considering and its acceptance criteria.
What I am concerned about, is if hipothethical allowable stress which would include membrane stresses and bending stresses will be not smaller than actual allowable stress stated in DNV for membrane stresses only...

 

[Crannog]

True, but the explicit definition of the membrane stress for an axial load (or global bending) is exclusive of the tubular geometry, hence that σ.x remains but σ.y isn't used (instead σ.h is used for the the hoop stress). C201 maintains the σ.xSd terminology but inconveniently doesn't define the stress, potentially as global membrane stress for a planer surface will include only (global) tension or compression but a tubular member will obviously see both.

I've never seen a specific allowable used for 'combined' equivalent stresses (over just regular equivalent stress) and cannot think of a logical reason there would be to have a higher allowable for direct than for combined axial stress - for a member, the difference of a stress from global axial loading and global axial bending is immaterial. I'm fairly sure that the membrane stress is inclusive of global bending.

If you are still concerned one could just email DNV direct and ask for clarification. My local contact is on vacation currently otherwise I'd be tempted to check myself.

D

 

[EngineerMickeyMouse]

Crannog, exactly ! I will have to formulate official query to DNV, nevertheless I would be pleased if you could ask your contact, once he got back from vacation.
The issue is certainly addresed to problem where bending stresses in 3D shell/plate element are not imposed by global bending, but local loads or in vicinity of investigated elements to which geometry is enforcing local bending, still not being global.