Qualifying a welding procedure for fatigue loads 3

[kingnero]

All WPQR's that are commonly in use, use(d) static destructive testing for accepting a certain weld.
The only exceptions I know are WPQR's according to EN 15594 and EN 16725, which are used for railway welding, and use in-track testing (and so endure a fatigue load by passing trains).

I know I can get fatigue tests done, but has anybody experience with describing such a test, including acceptance criteria, or could someone refer me to an existing standard regarding fatigue testing in combination with weld testing?

If, for example, you would do static testing of carbon steel to stainless using ER70-S6 wire, the weld would most likely pass using the standard destructive testing. The fatigue life would be much shorter than if 309LMo were used.
This is the general idea behind my question.

http://www.fusionpoint.be

http://be.linkedin.com/in/fusionpoint

 

[gtaw]

Interesting King.

AWS structural welding codes have no separate testing criteria for fatigue. They do have prescriptive design criteria that must be considered when cyclic loading produce stresses above a threshold value are exceeded.

Best regards - Al

 

[dhengr]

Kingnero:
I agree with Gtaw, I’m not aware of a specific fatigue test for welds; and AWS, AISC and AASHTO all have prescriptive design criteria for various structural/mechanical welded details and conditions as their geometry, etc. relate to fatigue life of that type of weld and detail. These various levels of fatigue resistance are based on years of testing and inspecting those various types of details. Obviously, you want a nice clean detail, without weld imperfections, and without nasty detail/stress flow geometry, or stress raisers, etc. etc. You certainly do not want a hard/brittle weld material or HAZ. I would err on the side of a compatible, but softer weld material (under matched) w.r.t. the base materials, at least not over matched and too hard, brittle or strong. Charpy tests would show toughness and that’s a step in the right direction. The way various test samples fail would give a good metallurgist (or good welding guy like Gtaw) direction for further improvement or recommendations for process and procedure improvements. The first S-N curve testing on most materials is just a rotating beam test. It can be done fairly quickly and economically. I would machine the rotating beams perpendicular to the line of the weld, with the weld at the center of the cylindrical beam, or at the max. moment location, depending upon how they do the test, and give em a spin. Some of this is just going to be good, conservative, experienced engineering judgement. You welding guys should have a feel for the welds (materials, process, procedure, pre and post heat, etc.) which give good elastic welds. Another direction is to talk with your welding materials suppliers. They should know some of this stuff about their products. Then, You get to determine when you think they are blowing smoke up your leg.

 

[kingnero]

@ gtaw: indeed, I'm using almost always European and ISO standards, but the general idea is the same. Eurocode three contains even a "fatigue atlas", with all kinds of details, and how it impacts the s/n curves.
But no "ready to go" prescriptions for what I'm looking for.

@ dhengr: There's a lot of truth in your statements.
From what the both of you are saying, the consensus is that what I'm looking for doesn't exist.
I have to convince a client to perform some testing, as I'm predicting trouble in the future, but it seems I also will have to make up the test and prescribe fail/acception criteria. This is out-of-scope, and will probably be seen by the client as an easy way to make money on his account. This is actually a whole project on itself, instead of being added to an existing contract.
I'll consult a local lab with experience in fatigue tests, perhaps they've done something similar in the past.

Thanks for your insights!

http://www.fusionpoint.be

http://be.linkedin.com/in/fusionpoint

 

[DekDee]

kingnero,
Very admirable what you are trying to do but personally I think you are wasting your time.
If it was the client/owner who were asking your questions I could possibly understand.
However, you have been contracted to supply a code-compliant weld (or structure) and that is what you should supply based on design criteria.
What fatigue loads the client/owner subjects your weld/structure to is of no concern to you.
Regards,
Shane

 

[gtaw]

Good morning King;

Fatigue is one of those subjects a client may not fully appreciate. One responsibility of an engineer is to make the client aware of potential problems and to incorporate the appropriate factors in the design.

As an engineer, there are design aids available to assist the design team. Rarely does the engineering team have the time or financing available to undertake basic research to expand basic knowledge on a subject as complex as fatigue. However, there are tools available such as fracture mechanics that allow the designer to predict the longevity of a structure subject to cyclic loads and subsequent fatigue failure. Generally, codes offer a prescriptive approach to designing for static as well as cyclic loads. However, the codes usually allow for advanced mathematical approaches as well. The advanced design tools available to the design team must be used with a full appreciation for the science behind the computerized tools. After all, the adage "garbage in, garbage out" still applies.

Dinosaurs like myself are limited to the prescriptive approach, while the young guns with strong computer skills can use the recently developed design tools. We sometimes question whether these young people really have an appreciation for the basic science behind the new tools. To that, I respond we have a responsibility to review their work to verify their approach is valid. With time and experience that tempers the blind faith of the young, the new tools will revolutionize our designs. In many cases, the computerized design tools have already changed the way we approach design. That being said, the advanced design tools are most likely to take into consideration things that are well understood and "easily" modeled such as geometry, notches, and other stress risers.

I think you have a valid question regarding the properties of welds between dissimilar metals. The qualification requirements of any code are a statement of the minimum requirements for methods of qualifying a WPS. As the engineer, you must consider whether additional testing and evaluation is warranted. I can see the potential for a welded joint between carbon steel and stainless steel suffering a degradation of the properties when subject to cyclic loads. That area I would be concerned with is the "mushy zone" of the weld where the carbon from the carbon or low alloy steel could alloy with the chrome from the stainless steel to form brittle chromium carbides. Their influence on the macro scale could go unnoticed, but under the influence of cyclic loads, it may be a different story. I would review the literature available from ASM to see if they have information available on the fatigue characteristics of dissimilar weld joints. There may also be information available from EPRI or government sources.

Your concerns probably are not addressed by codes such as the structural welding codes published by AWS where there may be an assumption the welded joints are between similar base metals. The problem is laid squarely in your lap to take any additional steps needed to ensure your design is safe.

Best regards - Al

 

[dhengr]

Kingnero & Gtaw:
This is a bit off topic for this thread but you guys might find it interesting. It relates to Gtaw’s excellent post of 19AUG17, 14:59, in thread725-429054, RE: “HAZ and NDT in Piping,” and his last post here, and the way small unforseen things can really change the picture. And then, to toughness, hardness, brittleness and low cycle fatigue in my case, as they might be issues in this thread too. We were making CJP welds in 1.5 & 2" thk. by about 10' wide in HSLA stl. plates, butt welds, many passes, about 1/3 from one side and 2/3 from the other side, SAW, with a flux with some alloying constituents in it. We had used this wire and flux combination for years without any problems. But, this time they really bit me in the butt.

I got an early a.m. call at home before even starting for the office. At least I didn’t have to go back home to pack. We had a large depressed center flat car sitting in a rail yard on the east coast, with a several million dollar transformer sitting on it, and a large crack in the deck pl., in the lower transition area of the deck pl. What to do, could I bring a new railcar along with me when I came out to have a look, can you get here by noon, etc? I sure didn’t have any answers during those first morning phone calls. On seeing it, the crack seemed to originate in the above mentioned CJP butt weld, and went toward the center of the car and also up the transition. Fortunately, the crack was in a pretty favorable orientation w.r.t. the stress fields, so I had them drill some holes through the deck pl. at the crack tips, did a bunch of calculatin, thinkin and prayin and then told them we wouldn’t pay for transferring the load (+ 200ton, near the car’s load limit) to another car, even if we could find one quick enough, and told them it would be o.k. to move the load to its destination. The transformer manufacturer, the handling railroads, the car owner, and the power company were in a real twit. They wanted the transformer moved safely to its destination, on time, and certainly didn’t want any of the responsibility if the railcar caused any problems. We did buy some extra short term insurance for this move and our ill situated involvement in it. We got regular reports on the car’s progress during the rest of the trip, as to any additional crack growth (none) or other unusual conditions (none), and then had the car brought back to our plant.

We took samples out of the weld area and the two base metal pls. for testing. The weld looked good when we sectioned it, and it had been UT’ed during construction. But, chemistry and hardness, etc. at different levels in the weld showed increasing manganese and hardness in the outer passes of the welds. These multi-pass welds were precipitating some manganese out of the previous weld passes and out of the flux too and concentrating it more and more in the outer passes, and the more-so with more passes. The crack did seem to originate in the outer passes of the weld. We gouged out the crack and repaired it, and we replaced the top layer, or so, of weld passes in the butt weld. The metallurgists told us to switch to a neutral flux and the same welding wire for the multi-multi-pass welds and that seemed to solve the problem. At least I’m not aware of any more cracks like that one, on a fair bunch of railcars of that general design. We did do some testing and inspections of other cars in this group. These cars were only a few years old, and when you consider the cars don’t regularly run under max. loading/max. stress (max. cap’y.) and some percentage of the time they are running with no load or only 3/4 max. loads, etc. so this was hardly a high cycle fatigue problem, and not even particularly highly stressed in the crack region. But, in this case the hardness/brittleness of the weld seemed to do us in.

 

[kingnero]

DekDee, gtaw, dhengr,
thanks to all of you for your replies.

@ DekDee: What fatigue loads the client/owner subjects your weld/structure to is of no concern to you
That's rather harsh. I am copntracted to do a certain job as a welding engineer, scope has changed (materials involved) but my initial assignment stays the same: making sure all welds are sound by design and by execution. I can simply write a note saying " I predict failure but you won't pay for testing, so all damage resulting is not my problem", but that isn't very client-friendly.

@ gtaw: The problem is laid squarely in your lap to take any additional steps needed to ensure your design is safe. That really sums it up. Now I need to propose a cost-efficient action plan, that takes the bull by the horns but in such a way neither the client or me gets harmed (financially or by reputation). Fatigue testing is an important factor here. As "standard" tests (with corresponding acceptance criteria) don't exist, it's hard for me to convince the client to get at least something done.

@dhengr, nice example. do you remember the basicity index of the initial flux?

http://www.fusionpoint.be

http://be.linkedin.com/in/fusionpoint

 

[DekDee]

Kingnero,
I think you have misunderstood my response.
You have stated yourself this is out of your scope.

"I have to convince a client to perform some testing, as I'm predicting trouble in the future, but it seems I also will have to make up the test and prescribe fail/acception criteria. This is out-of-scope, and will probably be seen by the client as an easy way to make money on his account. This is actually a whole project on itself, instead of being added to an existing contract."

It is the same as a Welding Engineer nominating B31.3 Normal Fluid Service for a piping system and the client then subjecting the system to Severe Cyclic or High Pressure loads.
Is the Welding Engineer responsible for that ? - of course not.

As far as I am aware the Design Engineer calculates the loads and nominates the required welding (and testing) - not the Welding Engineer.
The Welding Engineer ensures the welds meet the requirements of the designer.

Am I missing something here ?

Regards,
Shane

 

[dik]

An engineers job is to engineer... simply stated. If there is an issue, outside your scope, then you have an obligation to advise the client of such... maybe you can increase your scope... but, you should not complete your work with a potential problem without making the client complete aware of the problem.

Dik

 

[kingnero]

@ DekDee,
yes, there is scope creep, but no, that does not mean it's no longer my problem. There are now just additional problems that need adressing.
What fatigue loads the client/owner subjects your weld/structure to is of no concern to you.
If I tell them to go to hell, or have a lawyer write me a waiver as I stated before, the one guarantee I have is that I won't have that client any longer after this project.
So yes, it's my concern. A standard would be of tremendous help if it existed. As I said, I foresee problems, the owner only sees additional costs. I have at least to let him know what the actual situation is, and how to adress it. So that's why I came here for help.

@Dik,
yes, I'm trying to increase my scope, as this is for sure a very interesting problem. The biggest problem is that it's a quite enormous increase in man-hours here.

 

[dik]

king... comes with the project... and, quit whining about all the new work... <G> (Old text from FidoNet days, meaning a big grin)

Dik

 

[kingnero]

^^^ that ain't gonna do it.
However, truth be told, I'm considering doing it for reduced fee as I'll be learning quite a bit here as well. Might be an advantage for future projects.

http://www.fusionpoint.be

http://be.linkedin.com/in/fusionpoint

 

[weldstan]

kingnero
You can certainly perform cyclic loading tests on a simple welded specimen taken from a standard WPQR coupon. What is your desired result? Achieving 70% or some other value of the life of the base metal subject to the same loading parameters and how will you achieve same in the actual weldment? Minimizing metallurgical strength/toughness mismatches, residual stress, geometric differences and size of detectable weld flaws are keys to achieving fatigue life of the welded component close to that of the base metal. In a complex structure, it is doubtful that a very simple fatigue test would provide real predictable information regarding the actual life of the structure in a highly cyclically loaded environment.

You may be over engineering your problem.

 

[gtaw]

But, isn't it these types of problems and challenges that make our jobs so interesting to us?

Like King says, it can be a learning experience.

And Weldstan, you are hitting the nail squarely on the head. It is doubtful a single test will provide sufficient information to act as a predictor of the performance of the weld or the connection. It sounds like a multi-discipline approach would be worth considering.

Best regards - Al

 

[kingnero]

In a complex structure, it is doubtful that a very simple fatigue test would provide real predictable information regarding the actual life of the structure in a highly cyclically loaded environment.

I agree... that why I said:

This is actually a whole project on itself, instead of being added to an existing contract.

If anybody could offer any help regarding this subject, I'd be extremely grateful:
What is your desired result? Achieving 70% or some other value of the life of the base metal subject to the same loading parameters and how will you achieve same in the actual weldment? Minimizing metallurgical strength/toughness mismatches, residual stress, geometric differences and size of detectable weld flaws are keys to achieving fatigue life of the welded component close to that of the base metal.
Writing specs of a fatigue test shouldn't be too hard. Determining the acceptance criteria is, according to my humble opinion.

http://www.fusionpoint.be

http://be.linkedin.com/in/fusionpoint