ABAQUS FEA of Pressure Vessel - plastic collapse

[Chrissy123]

I have written an FEA model for a flat head pressure vessel,the head thickness is a lot less than the main body and it's a simple cylinder with a flat head, no valves etc. The FEA simulation is elastic-plastic; I believe I am meshing to a fine enough level as I have done convergence models etc .... I am assessing the collapse pressure by doing the following:
Ramping up the pressure by 1Mpa per time interval - up to a sensible max
Looking back at the ODB and for each step counting up the total no. of Gauss points that have plastically yielded - I then compare this to the total no. of Gauss points in the section of interest (which is the thinner head as this will fail first and I can see from the model where the max stress is).
For each iteration then as I'm summing up how many Gauss points have yielded I compare with total until I reach a percentage threshold (say 5%) and if this is reached then I will know from the ODB at what time increment this occurred at hence I can interpolate the pressure.
Can anyone critique (or otherwise) this approach ? I've yet to confirm from a theoretical point of view which I'm struggling with so if anyone can suggest that would be a help - thank you .

 

[DriveMeNuts]

I'm a bit confused. You are finding the collapse pressure using an elastic-plastic analysis. However you seem to suggest that you are not using a procedure that relies on convergence? Is that possible?

My second query is why are you counting Gauss points? Where does the 5% value come from?

Have you reviewed the method used in ASME VIII Div 2 Section 5.2.4? Why not just use that method?

Sorry for not being able to answer your questions. However, I am unable to understand your approach to your analysis.

Annex R in PD 5500 provides a simple but accurate method for calculating the secondary stress in the cylinder to head junction. This may be useful for finding an approximate of the collapse pressure.

 

[Chrissy123]

Thanks - I will take a look at what you have suggested. I have assumed that I can apply pressure to the model and then analyse the stresses and the plastic yielding, the 5% results tie in with the Yield stress for the material so I thought this was correct; I'm obviously making a big mistake here -

 

[TGS4]

You're definitely making a big mistake here. Use the procedure in VIII-2, Part 5.

However, you ought to be using the entire procedure, investigating all of the failure modes.

 

[Chrissy123]

Aah - I see so if I follow the approach ASME Section viii Division 2 standard for Pressure vessels, the limit for the hoop stress (away from any discontinuities) is that the material should not experience yielding. So this would be my criterion, when the first element reaches yield point the structure would be treated as if it has failed. Is that more like it ? Seems more simple than what I was trying to do anyway .... this piece of work is in part of learning how to use ABAQUS and Python scripting. Once I get more to grips with it all then I can start also looking at a more complex design piece but for now the model is a simple flat head cylindrical pressure vessel with no defects/cracks etc. So for starters if I investigate the failure mode as described as above then is that a good starting point ? I would appreciate your thoughts,
thanks,

 

[TGS4]

First of all, I have absolutely no idea where you obtained that assessment of the ASME Code. Seconds of all, that is completely wrong.

At this point, I will give to you the advice that I have given many many times over the last 15-20 years to engineers in your situation. You're in way over your head. You're completely unskilled and unqualified to be performing analyses such as this, and a such you are dangerous. Please stop. Please find a mentor to assist, take a training course, or otherwise correct your skill deficiency. For your work situation, please suggest that your company hire a consultant skilled and experienced in these matters. Hopefully you can learn something then.

 

[DriveMeNuts]

During my initial years of learning I was the only Engineer at my employer and therefore had no option but to teach myself.
I think if you jump straight into elastic plastic FEM without understanding the different failure modes, it is a certainty that you will become a dangerous "computer says yes" engineer who is oblivious to many hazards.
My advise would be for you to take a step back and learn about the elastic analysis method in ASME VIII Div 2. This will teach you how to interpret different types of stress. And the FEM analysis time is allot shorter.
Even before that, learn about the code rules and the fundamentals behind them. The code rules often calculate stresses and specify high stressed areas of interest. ASME Div 1 rules tend to be clunky and unintuitive. Allot of the PD 5500 and ASME Div 2 rules actually provide hints as to what the goal of the calculation is.
The best time saver would be to follow TGS4's advise and find a mentor. Unfortunately, often we work for small employers where this option is not available.

"when the first element reaches yield point the structure would be treated as if it has failed". This shows that you don't know allot. If this rule is applied to an element at the centre of a flat head, you will not be complying with code rules and the structure will definitely be dangerous. This is dangerous because the membrane plus bending at the centre of a flat head must be less than Yield/1.5. If this rule is applied to the sharp corner at the junction between head and cylinder, then this is very conservative because the stress can be as great as 2 x Yield. All of this only applies to plastic collapse failure. Plastic collapse is the best failure mode to learn about first, however you will then need to learn about fatigue, local failure, ratchetting and more before you can be considered a competent engineer.

 

[Chrissy123]

Hi to DriveMeNuts - thanks for that information - that is really helpful and yes I don't know a lot you are correct
I am a trainee and I am trying to learn and I don't have a mentor and I'm doing this to try to learn and just to assure (@TGS4) I will not be building an actual pressure vessel any time soon - nor advising anyone on the design as I know that I'm just starting out on this very long road of learning at the beginning of my career.

It's certainly a difficult and complex area, a lot more complex than originally thought and just learning ABAQUS in itself is difficult with all the different nuances so I was really just asking for advice.

As suggested I will start with plastic collapse failure - which is what I thought I had done by looking at individual nodes that had yielded but I've got this wrong so back to the drawing board - thanks

 

[DriveMeNuts]

I would start by re-familiarising yourself with how a principal stress affects the Tresca and Von Mises yield criteria. Wikipedia is good for this.
Then learn about the difference between Primary, Secondary and peak stresses and where (and when) to find them on a vessel. ASME VIII Div 2 provides this however it requires years of experience to fully decipher its meaning. As elastic stress analysis is not an exact science there are complex multiple meanings and interpretations of the categories of stress. If you can get your head around elastic stress categorisation then you will be able to interpret stresses in an elastic-plastic analysis more competently. As explained above, simply jumping directly to elastic plastic will result in you being oblivious to many hazards.
This can all be achieved with a second moment of area stress calculation sketch on a piece of paper. Using ANSYS is a fun waste of time for simple problems. ANSYS is more useful for complex geometry problems.