Is it possible to evaluate weld strength using FEA? 1

[Rhyder88]

Hello all
Is there any possibility\methodology to evaluate the strength of a weld using FEA softwares. Brief overview of the problem: A structure consists of sheet metal parts is failed when subjected to vibration test . The failure is observed because of welding. So is there possibility to evaluate this problem scenario in FEA??

 

[TGS4]

It depends. What is the actual failure mechanism - is it a strength issue or a fatigue issue? If fatigue, is it low-cycle or high-cycle?

 

[corus]

Given it's vibration and high cycle fatigue then there are design codes for the assessment of welds if you know the stress and the number of cycles. For this you could do a lengthy analysis of the vibration modes for a given dynamic movement of the base, or more simply just consider the range of maximum displacement that the structure sees at key positions and calculate the stresses from that. If you know the number of cycles then you can get the design stress range that the structure can withstand with a certain probability of survival, and assess your actual stresses against that limit.

 

[GregLocock]

More generally the answer is yes, because that is precisely what car companies do. BUT they have proprietary weld models that include the results of decades of testing and analysis.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[racookpe1978]

Seemingly trivial, but you do have to include it in your FEA model: The "as-welded" geoemtry has to relatively accurate to get any useful information: Fillet size, what cross-section shape, whar radii are assumed, what strengths are assumed at each part of the weld, exactly what part of the weld is "joined" and what part is "open" needs to be specific and accurate.

For example, you could model a simple butt-weld (double-sided full-penetration weld) easily across two identically-sized pipe walls.

But if you try to use that simple an approximation for a fillet weld across a socket joint, where thicknesses and radius and internal openings and the back side of the weld are not "melted together uniformly", you'll get useless (wrong!) results.

 

[Rhyder88]

Hi

The failure is due to fatigue(high cycle). which is the correct way to approach the problem? I tried to simulate using dynamic frequency response analysis. But the maximum stress is not near to the expected weld area.

 

[GregLocock]

Have you given the weld area the same material properties as the base material?

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Rhyder88]

I have simulated the weld using rigid elements. With base material property

 

[corus]

Failure at welds isn't necessarily in the weld material but more commonly at the toe of the weld, in the parent material, nor is it necessarily at the maximum stress as this may occur in the parent material away from welds for instance. Failure may be due to a relative low stress at the weld, but high cycles. For stresses at the weld take the nominal stress, ignoring any peak stresses from the weld geometry or from your rigid attachment. You can do that by looking at the stress distribution as stresses approach the weld. Also look at the direction of stress in relation to the weld as this may give a different weld classification and a worse fatigue life.

 

[ahmedi811]

Well, you need to model the weld either as a solid or shell element. (Chalmers also recommends increased thickness). Also you need to consider the following:

1. Weld residual stresses caused by welding.
2. Weld toe dimensions as recommended by AWS specifications for groove welds.
3. Introduction of hot spot stress extrapolation points as recommended by DNV or many research papers related to FEM of welds.
4. Provide appropriate weld material properties.
5. If you like to opt for notch stress method, in that case you need to model a crack into your weld model and investigate the HSS.

HSS will be nothing but the weld strength. However, the method of loading and connection details determines whether your weld strength is in terms of shear or bending. Generally, groove welds resist bending while fillet welds resist shear.

Nevertheless, there are few points that still require answers from dear readers:
1. How many points are recommended to be introduced into the FEM Model around the periphery of a Tubular K-Joint for determination of Hot Spot Stresses?
2. If no weld is modelled at the joint. i.e. only the two extrapolation points are, would that affect the result for HSS?

 

[ahmedi811]

A small correction: HSS is NOT equal to weld strength. However, in case of fatigue, HSS can be used to determine SCF and using appropriate S-N Curve, you may obtain the number of cycles to failure. Apologize for that mistake.

 

[Rhyder88]

Thanks Ahmed.. could you please give reupload the attachment. The link is not working.

 

[ahmedi811]

That wasn't an attachment. Another mistake by me. Apologies once again.

 

[corus]

Ahmedi811's method may be a definitive method of assessing welds but there's no real need to go into such detail as the SN curves provided in design codes already take these factors into account. Their curves are based on actual measurements, and take a certain probability of survival from their data. British, Swedish, and possibly German standards have different SN curves for differing weld classifications whereas ASME, if I recall, uses a standard SN curve with varios factors to be applied to the stress depending on the weld geometry and load. The two methods amount to the same thing at the end of the day.