Weld joint as per ASME is failing (FEA Analysis) at working pressure when bolt preload is applied

[Gokulkrishna]

Hi,

I'm new to FEA analysis. We design our valves considering wall thickness from B31.3 and the weld bevel dimensions as per B16.25.

Our valves are not failing at hydrotest pressure (1.5 * working pressure) during actual testing. But while performing FEA on the valve the weld joint is showing results that are to close to the yeild strength of the material at the working pressure itself which means that our weld joint will yield at hydrotest pressure. Which is not a good scenario.

Can someone help on this. We are considering the entire valve assembly with the valve connected in the pipeline with studs and nuts and pretension is applied on the studs. Also we have not considered the gasket between the two flanges as the gasket are spril wound and is a consists of 2-3 material(hence difficult to model and also mesh).

The weld joints are safe as ASME standard.

Thanks in advance.

 

[desertfox]

Hi

Well I am not familiar ASME in depth but I would say if you've checked your weld stresses independently of FEA to the above code and found them to be safe, then I would think there is a mistake in your FEA. Try using the test pressure in your hand calculations and see what stress that produces, if the result is the welds are okay then you definitely have a problem with the FEA.

 

[TGS4]

You say that the welds in your analysis are "failing" because they have stresses that approach or exceed yield. What exactly is your criteria? Try following the guidance provided in VIII-2 Part 5.

 

[dhengr]

Gokulkrishna:
Welds are notoriously difficult to model properly in most FEA software. You just don’t/can’t mesh things small enough or properly to really read the detail. They and their details often show hot spots which are usually not a problem (I didn’t say never a problem) in the real world. Around the weld you have many reentrant corners, shape and size/thickness changes, changes in direction of stress flow, etc. etc. These kinds of anomalies often show up as high stress regions. However, if they are well confined by surrounding material at much lower stress, they usually aren’t a problem. At most, you will likely get a little yielding at the hot spot to slightly redistribute the stresses, but the lower stressed adjacent area will pick this up with little trouble. You can try to ease the shape change or geometry at these locations, but the biggest thing is that you not introduce new/worse stress raisers at these locations with any inferior welding (porosity, undercuts, lack of fusion, etc. etc.) or inferior weld details.

 

[weldstan]

Have you taken into account the generally, higher yield strength of the deposited weld metal vs that of the base metal?

 

[jte]

Stop... Listen to the sage thoughts of TGS4.

"Thou shalt not yield!" may make for a good line in a movie, but given a linear elastic analysis would be downright foolish as an acceptance criteria.

 

[Gokulkrishna]

Hi All,

I did some modifications to my FEM.

Earlier we were not using a gasket in the assembly, hence the stresses from the bolt preload directly got transferred into the weld area causing it to go beyond the yeild value.

Also we are considering that the after welding the weld area will have the same properties as that of the remaining body. This way the analysis bacame relatively easier.

So two analysis were performed:
1. with gasket and bolt preload to connect the valve to the pipeline.
2. only the valve was analysed.

The weld area is now on the safer side. But the issue is now near a hole (for a threadolet) at 60deg from the horizontal axis. The maximum stresses are observed on the edge in the inner side of the meter. Please see attached snap (area highlighted in red). My guess is that since the edge area on the left side in the attached snap has relatively lesser thickness and has the pressure applied from both sides (i.e. on the ID of the valve and the hollow area for the threadolet) the stresses are relatively high. Please let me know if I'm missing out on something else. I'm not that well conversent with the ASME BPV codes.

Thanks & Regards,
Gokulkrishna Goli

 

[tbuelna]

As dhengr noted, it is difficult to get an accurate result from FEA of an assembly. While your assembly may have passed a pressure test, you must also consider what are the full requirements for qualifying the design. Usually, designs are certified by analysis and the qual test is primarily used to validate the analysis.

The reason for this is that a proper analysis will take into account all of the factors like stress concentrations in weld joints, minimum mechanical properties in materials, worst case combinations of loads, etc, while the qual test of a single test article may not.

So the question for you is what is the certification requirement for your design? Is it required to be certified by analysis, by test, or by a combination of both?

 

[TGS4]

At this point, because you, as you say, are

not that well conversent (sic) with the ASME BPV codes.

I'm going to go ahead and say that you are unqualified to be performing such an evaluation. It is not merely a matter of whether or not elastically-calculated stresses exceed yield (or don't). There are multiple failure modes that need to be considered, along with the considerations mentioned by others in this post.

This starts with defining your Code of Construction. If it is ASME B31.3, then you need to ascertain whether or not you need to be performing any design by analysis to begin with. If you are, then B31.3 directs you to use ASME Section VIII, Division 2, Part 5 for Design By Analysis. Therefore, you are obligated to use it - ignorance is not an excuse anymore.

Please, determine if a design by analysis is actually required in your case - without making any modifications. Especially if your assembly includes a bolted flange joint, what you are doing could get someone killed. Then, hire someone who is well skilled and well experienced in design by analysis to the ASME boiler, pressure vessel and piping Codes.