In general, and I expect other could add to this, assuming a given classification of carbon steel filler metal, and assuming the weld reinforcement is ground smooth, fatigue performance of weld metal should be improved with a finer transformed microstructure, should be improved with less hard transformed phase in the microstructure, should be improved with finer second phase inclusions, should be improved with inclusions having more spherical morphology, should be improved with using "cleaner" filler metal with lower analyzed residual contaminant elements, and should be improved with less and smaller weld discontinuities from the welding process.
Reading between the lines to what this means in practice is correctly calculated bead size and controlled heat input for the given weld process, and using chemically tested and verified "clean" consumables. In the absence of a defensible fracture mechanics basis for maximum flaw size based on flaw type, the welding should be conducted such that no discontinuities are indicated within the confident resolution of the NDE methods used. In general, for a given heat input, carbon steel filler metal with increasing alloy will tend to give a finer transformed structure (reduced proeutectoid ferrite relative to acicular ferrite). Carbon should be controlled as to as low as practical for the application.
It should be borne in mind that no general discussion of weldment fatigue life will substitute for actual fatigue testing in critical or complex applications.
I've looked at a number of small bore piping socket weld failures in our plants, and as others have mentioned, the failure initates at the toe of the weld in the base metal. I have not seen a vibration fatigue failure that initiated in the weld. So, changes in weld metal composition or properties would have no affect on this failure location. For our failures, vibration fatigue is most always the cause. The types of changes that improved fatigue life were changes in geometry of the fillet, or changes in support of the piping to reduce the vibration amplitude. Concave weld profiles, blending the weld toes smoothly to the base metal and a longer fillet leg on the pipe side (2 x 1 fillet profile) are the usual geometry improvements that work.
Having said all that, there are other fatigue mechanisms where improved weld metal fatigue resistance would be helpful. On boiler waterwall panels there is a membrane welded between the tubes in the furnace. The mebrane is a flat bar welded to the tubes with from both sides of the joint to achieve complete penetration welds. If complete penetration is not achieved, the unwelded portion in the center acts as a stress riser for mechanical fatigue due to heatup and cooldown of the boiler during startup and shutdown. Fatigue intiates at the root of the weld and progresses either through the weld or through the tube. So your fatigue test should have a notch or precrack in the weld metal to measure resistance of weld metal to fatigue crack propagation. If you just put a fillet welded sample on a vibration table with no notch or precrack, it will likely always fail in the base metal and you won't be measuring weld metal fatigue resistance.
Thank you for the info, very interesting. What type(s) of material(s) are you using in the boiler and which welding process?
For what it's worth...alot of small bore socket weld failure can be contributed to a lack of gap at the time of fit-up of socket weld joints. It is the industrial standard to bottom out the pipe (in the socket) and then raise the pipe 1/8" to a minimum of 1/16" and then applying the first tack. If this is not done, then eventually (due to expansion & contraction) of say, steam & condensate lines fatigue failure will occur because if the piping is bottomed out in the socket....the pipe itself in this contraction/expansion phase will fatigue the weld to failure no matter which weld process is used or how much integrity is in the welded joint.
I'm sorry, but I don't agree that a lack of gap in socket welds contributes to fatigue failures. This is similar to an urban legend. If the pipe and socket are the same material, they have the same thermal expansion coefficient and will expand at the same rate as the piping heats up and cools down. Therefore, there can be no cyclic stresses and no fatigue. You are correct that it is an industrial standard to require a minimum pullback prior to welding. As best anyone can determine, there was no technical basis for establishing the requirement, only a concern that it might be an issue. However, there has been research by EPRI and others to show that a lack of gap does not decrease fatigue life.
Aside from the gap and fatigue issues while piping is in service, isn't the typical socket weld fit-up sequence just a precaution, during fabrication, to prevent the pipe from expanding against the fitting and cracking the fillet weld?
That was my point - if the socket and pipe both expand at the same rate as heat is applied, how could it crack the weld? A person could weld socket joints all day long with the pipe bottomed out and they would not crack during welding or during operation due to thermal expansion/contraction. I would be interested in hearing about any firsthand socket weld cracking experience that was due to thermal expansion/contraction of the pipe inside the socket.
I agree with you with regards to the expansion or contraction of the weldment while it is in service and undergoing gradual changes in temperature. I'm questioning the effect of the rapid change in temperature during the welding process only and the idea that the pipe expands faster that the fitting.
Why would the pipe expand faster than the fitting during welding if the thermal expansion coefficients are the same?
I love this sort of thought experiment. The only way I can think of to make the pipe expand more than the fitting would be to run very hot fluid suddenly into a cold pipe. Since the pipe is inside the fitting it will be heated first. The little spigot inside the fitting may even be insulated by whatever radial gap exists between the pipe and the fitting and this might cause an even higher temperature difference. I have no idea how much difference could be expected. If the weld is sufficient to develop the strength of the pipe I would expect the hotter and thus weaker pipe to upset on the end rather than the weld to fail. If the shoulder in the fitting is smaller than the wall thickness of the pipe this would be practically guaranteed.
I would be surprised to find that analysis of the distortion produced by welding does not indicate that the spigot will shrink back enough that further contact is virtually impossible.
Still if the code, or even common practice, is to leave a gap it's easy enough to do.
Bill
I too appreciate the opportunity for this kind of exchange here in the forum. The term "faster" may have been a poor selection on my part. I am visualizing a weld puddle temperature capable of melting the base materials in one location (the joint), and lower material temperatures a short distance away (the body of the fitting and the pipe). Due to the difference in shape and mass of each item there may be a brief unequal absorption of heat by both. Considering that expansion/contraction occurs in all directions, I'm thinking that uniform radial expansion of the pipe is restricted by the fitting forcing an abnormal or non-uniform expansion of the pipe in the axial direction. Once welding has stopped, uniform contraction would be influenced, to some degree, by the fillet weld.
Yes, a very interesting topic and I appreciate the discussion. I did some more digging and found some research that shows when the torch is directed at the root of the joint and the puddle doesn't favor either the pipe or the socket, the heat distribution is fairly equal. When the puddle favors the pipe, the pipe does absorb more heat and could expand faster than the socket in the lengthwise (axial) direction. The research data included both test samples and finite element stress analysis with zero root gap and also suggests that solidification or hot cracking can occur at the root of the fillet due to the thermal stresses in carbon steel materials. If a root crack were present prior to placing the pipe in service, I could see how it might propagate as fatigue.
If anyone has access to the Elsevier Science Direct database, here is the reference:
Study on crack generation at root of socket welds.
K. Iida , F. Matsuda , M. Sato , M. Nayama and N. Akitomo.
Nuclear Engineering and Design, Volume 166, Issue 1, 1 October 1996, Pages 85-98.
I agree that you could weld (bottomed out) socket welds all day with out ever having a problem, but in the service of the joint where expansion /contraction is prevelent, then it is likely the weld would be exposed to an increase likelyhood of fatigue failure of the fillet weld because of this (expansion/contraction) senario.Stress is increase where the pipe is bottomed out as it acts like a fulcrum and stress on the legs of the fillet are increased due to the temperature changes incrued by this joint design.
If this was an "urban legend" then why are there spacer (rings) available to prevent the socket (pipe) from bottomimg out???
The point I am attemping to make is that in the service of the weld,
eventually joint failure would be more likely then if the joint was made up the way of the indusrial norm (i.e. gap in the socket).
Contraction rings are an alternative to the scribe & pull out or cock & tack methods to insure code compliance for the minimum 1/16" gap prior to welding. From a marketing point of view, fit-up time is shortened when using contraction rings.
"If this was an "urban legend" then why are there spacer (rings) available to prevent the socket (pipe) from bottomimg out???"
The answer is: Because someone found a way to make money off of a Code requirement. The same way they found a way to make money off of fillet weld gages, hi-lo gages, skewed fillet gages, and other inspection tools meant to check dimensions required by codes and standards. My feeling is that just because someone found a way to make money off of a code requirement, doesn't mean the code requirement had a sound technical basis.
My understanding has always been that B31.3 called for a gap on socket welds to accomodate weld shrinkage.
What I've heard tends to agree with MBSims that recent testing indicates that the gap does no good.
Until the code changes it doesn't matter in a practical sense tho.
JTMcC.
BWeldor,
Thank you for your time and intellect.
I agree with all of your concepts, I am not trying to challenge academics. :)
What experience do you have or had with these types of weld joints? Did you every try to use an alternate filler metal or did you always redesign or develop the components/structures?
I am very interested in personal experiences.
What are you manufacturing? This post smells a little fishy because you seem reluctant to specify design conditions.
Please elaborate.
It is easy for a toPic to drift. I don't reAlly see how this topic is fishy? It seems pRetty strAight forward to me.
I just waNted to focus On filler metals and maybe process vs. design. My focus Is on welding, not Design.
We are manufacturing equipment to vibrate material for separation, as stated in my opening/ original paragraph.
:)
So if I read "PARANOID" in the text does that mean that I am, because I looked for the reason that you mixed in upper case characters?
I can understand vonash's concerns, although I didn't see any problem with the posts.
We have all seen some posts on state-of-the-art-armor-plate welding that smelled fishy at the time. Mostly because the poster seemed to originate from Malaysia and would not clarify his/her questions. It seemed odd at the time and imaginations can run wild given the state of world events. None of us would want to inadvertently contribute to terrorist attacks.
At any rate, hopefully you got the information you were seeking. I don't recall all the responses due to the length of this thread, so I hope this is not a repeat. But I was going to suggest Lincoln Electrics publications on welding for fatigue situations. Some of that is on the web and other information is available through Lincoln's reasonably priced publications.
My original comment on preheat assumed you were dealing with thick materials. You clarified that the material was not.
Chet Guilford