Local peak stresses in solid models

[Struct71]

Hi,

I am regularly doing analyses with SOLID45 elements and plastic material properties with TB,KINH defined.

In many models I find very local peak stresses and strains. These are higher than allowable according to classic design codes (yield/1.5, tensile/2.35, yield*0.6). These codes do not allow for plasticity in the first place.

While I am sure that the stresses I find are local and no problem to the structural integrity of the model, I am having difficulty to relate the stresses to codes. Statements like "The red elements are OK because the element right next to it is green" are very slippery and not satisfactory.

Does anyone know a way to evaluate local stresses around yield stress? Is there a code which is more suitable for FEM analyses? I am thinking of a statement like "peak stresses only occur on 5% of the load bearing surface".

Thanks,

 

[TGS4]

Which design codes are you referring to?

 

[gurmeet2003]

eel071,

I do not think there are any codes of the type you are looking for. Either use the codes for design or use the analysis. The proof for the analysis has to be provided by
testing or comes from the experience of the analyst.

Gurmeet

 

[Struct71]

The codes in this case are PD5500 (tensile/2.35 or yield/1.5) and AISC-ASD (yield*0.6 for axial stress).
In fact they come to the same thing - no plasticity allowed.

The field I am in, which is somewhat conservative, I do not only need to convince my self, I also have to convince colleagues, third parties, warranty surveyors and insurance companies. It would therefore be nice to be able to say "occurring value is less than allowable value, therefore it's OK". Just waiving away high local stresses as not important will not fly in my world, however much I would like to.

 

[TGS4]

You're going to get yourself into a whole world of problems if you do not UNDERSTAND the intent behind the pressure vessel code of PD5500. The allowable stresses that you refer to are typically referred to as general membrane values. I am much more familiar with ASME Section VIII, Division 2 design-by-analysis rules, so the remainder of my answer will draw from that...

The allowable stress (based on the lesser of yield/1.5 or tensile/2.35) is to provide a suitable margin against plastic collapse and "burst" of a general pressure vessel. The localized stresses, however, have different, shall we say, allowables. For example, the local membrane stress may be allowed to reach 1.5 times the basic allowable, while the local membrane-plus-bending stress may be allowed to reach 3 times the basic allowable stress (all of this assuming that you have performed an elastic analysis). If, OTOH, you are performing an elastic-plastic analysis, the rules again are completely different.

Suffice it to say, this is not something where you can just "run" an FEA on a problem, say the local stresses at-a-point are below some value, and be done...

 

[TGS4]

One other point - SOLID45 would not be my element type of choice for a pressure vessel. SHELL93 would be my first choice for a thin-shell vessel (D/t>8-10). Otherwise, I would use SOLID95 - quadratic elements will give you better grid-convergence with fewer elements/nodes. But, you would know that because you are performing a grid-convergence study anyway, right?

And, your model doesn't have any numerical discontinuities either, because the stresses at those locations are identically infinite, so...

 

[MBlackman]

Also check out the SOLID 186 20-node element. You should use two elements in the through thickness direction.

There is a WRC bulletin 429 that recommends that it is inappropriate to consider calculated stresses in sharp transition regions, as for one thing it is considered very unlikely that ratcheting could occur there.

One method to estimate a stresse at a junction which is more in keeping with the intent of the pressure vessel codes is to obtain stresses at say 2t, 1t, 0.4t and use a quadratic extrapolation to estimate a stress at the junction.

 

[Struct71]

Some general reactions: The item I am evaluating is a modified thick walled pipe (OD18in/WT32mm) with holes, so D/t=22, which can't really be considered thin walled. I am using 4 elements SOLID45 in thickness.

The element solution PLES is the same as the nodal solution PLNS, and I have tried mesh refinement but there is no significant change.

The load is not pressure but axial load and bending, so the application of pressure vessel code is doubtfull. If PD5500 states that 3xbasic allowable is allowed for local stress, this will be 2x yield with elastic material properties, correct? How do you define local stress? When does local stress become global stress?

Thanks for your help.

 

[corus]

I'm not sure about PD5500, but BS5500 (or the european equivalent) pressure vessel code gives descriptions of allowable stresses which could be applied to any structure in my book. Your local stresses would be classed as membrane + secondary + peak stresses (basically the total stress) and would occur at a stress concentration, for example. If it was at a strucutral discontinuity (a stress necessary to satisfy structural continuity) then it may be classed as secondary (strain dependent). They give examples of locations of these stresses for classification. ASME VIII has similar descriptions. For your stresses at a hole then I'd class that as a stress concentration and as such would be a peak stress. These elastically calculated stresses have no limits other than those given for fatigue assessment for the expected number of cycles within the required life. You need to look at the stress distribution up to the edge of the hole so you can remove the peak component of the stress concentration. For a perforated shell the remaining stresses within the ligaments are classed as primary membrane plus bending with a limit of yield for bending. If it was an isolated ligament between holes then the limit is twice yield for that secondary stress.

corus

 

[Struct71]

Thanks All,

What section in PD5500 or ASME VIII indicates the use of local stresses?

Regards

 

[TGS4]

If your are looking to ASME Section VIII, then I would point you to ASME Section VIII, Division 2, Part 5. Be fore-warned, however, that if you are doing this for a pressure vessel application, that you will need to satisfy ALL of the applicable requirements.

You still have not yet told us what is the (legally mandated) Code of Construction for this "thick"-walled pipe with holes that you are designing.

(As a further aside - D/t>8-10 can be considered to be thin-wall - Reference Timoshenko's Theory of Thin-Wall Structures).