Fillet weld full penetration for plate in vibration fatigue? 3

[homebrewdude]

I have a 1/4" plate that is welded to a 1/2" plate.

The 1/4" plate is subjected to a vibration type fatigue load.

The current weld is 1/4" fillet all around the 1/4" plate.
Our supplier is actually oversizing the weld and is looks to be about 5/16"
Both plates are ASTM A572 Grade 50

Looking at the plate, it seems that the fillet welds completly penetrated the 1/4" plate and are both touching each other. (if looking at a cross section of the plate)
It also seems like the heat from the weld undercut the material.

The 1/4" plate is failing at the weld due to vibration.

Is it normal practice to put 1/4" fillet welds on both sides of a 1/4" plate?
Is the supplier just using excessive heat?
Could the failure problem be due to the weld area being brittle? As all the vibration is being transfered through the weld.

Would a better weld be bevel groove type?
I would then have to pre-chamfer the 1/4" plate?

 

[metengr]

Where exactly are the weld failures - along the edge or toe of the fillet weld or elsewhere? The fillet weld is normally sized base on the thinner of the two base metals being joined, in your case the 1/4" fillet sounds right. Welding from both sides is acceptable and would increase your load carrying capacity. Unfortunately, fillet welds exhibit poor fatigue properties.

The point you made regarding undercut is a very important observation. If the failures are occurring along the edge of the weld where the undercut is prevalent, one method to increase fatigue resistance is to simply blend grind the weld toes. Weld toe undercut can act as a stress riser, and induce fatigue crack initiation and propagation.

Before getting into any further recommendations, I would blend grind the weld toes and see if this helps. If not, your second option is to bevel the 1/4" plate and use a full penetration weld with a small fillet to reduce stress concentration.

 

[homebrewdude]

I have not seen the failures, as the 1/4" plate is completly lost in the failure.

Is it normal to have an undercut with this type of weld?
Or is it being overheated?

 

[metengr]

Undercut is a function of weld technique, and it should be prevented or removed. I looked up the A 572 specification and it is basically a low alloy structural steel, with the Grade 50 being 50 Ksi minimum yield strength. Typically, the Vanadium and Columbium additions help with controlling grain size for uniform properties.

I can't provide any comment regarding overheating because you need to evaluate the location of failure, and welding process that was used. Also, one would need to know the specified heat treatment condition of the plates.

One comment that I could make is to be sure that low hydrogen electrodes are specified for welding. For this material and using the SMAW process, I would spec out E7018 H4 weld electrodes.

 

[homebrewdude]

And if they are not using low hydrogen electrodes, could I have hydrogen embrittlement.

They are not using any pre or post heating on the parts.

 

[metengr]

The base metal thicknesses you stated do not warrant the application of supplemental preheat provided welding is performed at ambient temperature.

Could it be hydrogen embrittlement? Normally, you have transverse cracks (cracks that run perpendicular to the length of the fillet weld) that form in the weld deposit from hydrogen embrittlement (aka delayed cracking). Again, you would need to evaluate the location of weld failure. I always recommend low hydrogen electrodes for welding any structural shapes - it is good engineering practice.

 

[homebrewdude]

I guess going back to my original question.

The 1/4" plate is being welded on both sides with welds that are oversized. We call out 1/4" but I measured some and they are as large as 3/8"

Since the welds are oversize, won't this always cause an undercut condition?

 

[metengr]

Correct. Oversize fillet welds will have an adverse affect on fatigue properties. Oversizing is not recommended.

 

[homebrewdude]

Assuming my part does not see any loading (just vibration)

I can probably actualy undersize the welds..

 

[pipewelder1999]

I may be wrong here but wouldn't the bead shape and contour have a great deal to do with the resistance to fatigue type failures.

Wouldn't Excessive convexity, overlap, abrupt changes in weld size and similar discontinuites reduce fatigue life more than an oversize weld (provided the oversize weld is free of these discontinuities).

I have read some articles regarding weld toe contouring that seemd to indicate additional passes or grinding could increase fatigue life.



Gerald Austin
Iuka, Mississippi

http://www.weldinginspectionsvcs.com

 

[metengr]

pipewelder1999;
Yes, regarding bead shape and abrupt changes as you stated because these are in effect local stress risers. But do not underestimate the adverse effects of increasing fillet weld size, thinking that this will increase load carrying capacity. I have seen fatigue test data showing the effects of added weld cap reinforcement applied to groove welds and increased fillet weld size, with both showing a noticeable reduction in fatigue life.

 

[pipewelder1999]

Is that related to the size or the added residual stress from the extra welding ? I do agree about avoiding the "more is better" thinking.

Thanks


Gerald Austin
Iuka, Mississippi

http://www.weldinginspectionsvcs.com

 

[metengr]

I believe it has more to do with geometry, and stress concentration. The reason I say this is because this reduction in fatigue life would also apply to post weld heated welds.

 

[unclesyd]

Adding one additional point on the weld contour. The closer you can get to a 3:1 elispse in the weld contour the better off you will be. The long leg should be on the 1/4" plate.

Absolutely no undercut on the 1/4" side and if there is any on the 1/2" side it should be taken out with a burr so the contour at the toe will rounded and grinding striations perpendicular to the weld. Grinding on the 1/4" side should be accomplished with a small wheel so the grinding striations will be perpendicular to the weld.

Don't forget the ends of the 1/4" plate, very important.

 

[unclesyd]

Forgot to add.

A cold lap of weld metal on the 1/2" plate is about as bad as undercut.

I would NDT test the weld with Magnetic Particle or Dye Penetrant prior to putting the weld in service.

 

[HgTX]

Even the designed 1/4" welds could be considered oversized, since the two throats add up to more than the thickness of your 1/4" plate.

Where exactly is the weld failing? Are the welds made one side at a time or simultaneously?

Hg

Eng-Tips guidelines: faq731-376

 

[stanweld]

homebrewdude,
Assuming that your design is new, I would suggest redesign using full penetration groove welds with reinforcing fillets. As metengr has so rightly stated, fillet weld designs are notoriously poor in fatigue service and can be as low as 15% of the base metal's fatigue strength due to geometry and weld residual stress. There are built in notches at the weld toes and root even with the best workmanship. On the other hand if your design is old and proven, then I would assure that the completed fillet weld geometry meets the conditions so ably described by unclesyd and metengr.

 

[BillPSU]

homebrewdude
What kind of loading is the joint experiencing? What kind of frequency is the vibration? How long before failure?

Most weldments are designed using static loads and the use of 1/4" fillets all around are very common. The weld strength will far exceed the plate strength if properly applied(proper filler metal, no undercut, no porosity, proper fillet shape and size). The problem you are having is not a static load but a dynamic load being applied many times. This vibrational loading may be causing stress concentration due to the direction of the loading and the type of loading. Redesign of the joint to spread the loading may be required. Your welding practice may never solve your problem.

FEA and strain gaging can give you good data for the approving the initial design of a weldment however life cycle testing in my opinion should be done whenever possible.

 

[elpoblano]

homebrewdude,

One issue is puzzling me, doesn't the maximum fillet size on a 1/4" or greater base metal shall be 1/16" smaller than the thinner thickness joined? You have a 1/4" and a 1/2" plates, shouldn't your maximum welding size be 3/16"?

Metngr and standweld have very solid statements, I would redesign the joint based on dynamic loading.

 

[homebrewdude]

WOW!! so many replies to this thread.

I have not personally seen the failures, and they are not very common.

This is an existing product, but I am working on a new product and do not want to carry over any problems.

I agree about the weld size.
We call out 1/4" weld, but actual weld size is closer to 5/16"~3/8"

The weld looks like it is placed in two passes
bottom left then top right.

The vibration is unknow, I am guessing 30-150 hz up to 10 g's

The only loading is the weight of the welded on 1/4" thick piece. 1/4 lb.