Residual Stress in (Shop) Cold Bends 1

[auzie5]

My question is: Does removing a coupon from a test bend relieve the residual stress in that coupon? Or can I remove a small coupon from the test bend and still accurately measure the residual stress introduced into the steel during the cold bending process?

Overview:

I am currently exploring bending large bore steel line pipe using a hydraulic ram bending machine. As I was looking for an instrument to measure the residual stress left in the pipe after bending I came across a product (Restan MTS3000 – see video demonstration:

https://www.youtube.com/watch?v=1ursJRGGFMI).

Since we complete destructive testing on a test bend before moving into production, I liked the idea of following ASTM E837-13 to determine residual stress by the hole-drilling strain gage method.

However, I am uncertain about the effects on the residual stress in the pipe if I remove a coupon from the test bend for testing in the Restan MTS3000 instrument.

Background:

I am in the process of cold bending some NPS 20 x 12.5mmWT, CSA Z245.1 Gr. 448 line pipe to a 25D bend radius and 30 degree bend angle using a hydraulic ram bending machine. As one of my checks, I would like to demonstrate the bending process has not introduced unacceptable levels of residual stress into the finished bend. However, I am unclear if I can remove a coupon from the finished bend to perform ASTM E837-13 tests or if I would be required to mount the Restan MTS3000 on the finished bend since removed a coupon may relieve some of the residual stress in the steel after the bending process.

 

[weldstan]

The cold work strain will exist in the coupon and the pipe bend; however, on a 25D bend, very little cold strain will have occurred. I truly would not consider over analyzing.

 

[auzie5]

Thanks weldstan. In actual fact, it is very unlikely I would ever actually perform this test. However, when we began considering cold versus induction bends for this application, some colleuges had reservations of using cold bends due to residual stresses left in the bends. I have never heard of anyone measuring residual stress in cold bends (especially ones this big) but I figured I would at least go through the thought process of how I could demonstrate that they are not a concern.

I appreciate your input.

 

[racookpe1978]

Well, by definition, when you form the bend (5 deg, 15 deg, or 90 degree) the metal has been pushed (pulled) past its yield point through the plastic zone into its new shape. When the pipe is released from its furthest bend position, it usually relaxes (stretches back) about that 2-3% into the actual final position.

The Codes expect this relaxation, and "expect" the remaining stresses in the pipe walls.

 

[LittleInch]

Your first question, of course it does.

"Residual stress" is a strange concept to try and measure in a cold bend though. Once you've bent it then taken the force away, what stress is left?

Sounds to me like you're trying to probe a negative and that is always difficult.

I suspect your colleagues don't result understand what a cold bend is?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[auzie5]

Thanks for the input racookpe1978 and LittleInch.

LittleInch,

I tend to run into people who prefer induction bends over cold bends as a rule. As soon as they hear 20" they immediately conclude it must be fabricated using the induction bending process before I've even had a chance to detail the bend radius (example: 25D).

I do admit that the idea of residual stresses left in a finished cold bend is a bit confusing to me at the moment. Is the general statement that, "cold bends have more residual stress left in them than induction bends" accurate? I would assume that depending on the quenching rate, induction bends can have residual stresses left in them as well.

For cold bends, I have always just completed my mechanical testing on my test bend to qualify the cold bend procedure without giving any serious thought to residual stresses existing in the finished bend. Unless for sour service, I cannot remember an instance when I post bend heat treated a cold bend (even for much tighter bend radii). I have always thought that if there are residual stresses left in the bend, they would impact the results of my test bend mechanical test results. Thus if I passed my mechancial tests on the test bend I would not need to worry about residual stresses. However, tensile tests are typically from transverse coupons to qualify the bend procedure. Should I maybe consider pulling some longitudinal samples as well to check if any residual stresses are impacting longitudinal properties?

 

[auzie5]

LittleInch,

Just to clarify, are you saying removing the coupon from the bend relieves the residual stress left in the cold bend (i.e. you cannot remove a coupon and measure residual stress? You need to measure residual stress on the intact bend?)?

 

[BigInch]

It is a little like measuring the position of an electron. Once you touch it, its no longer the same electron it was previously.

 

[chicopee]

Once you have determined the strain within the bend, why not get a copy of the stress-strain curve from which you can estimate the residual stress; of course, there will be a relaxation of the stress to e taken into account after you did the bending.

 

[LittleInch]

What I'm saying is that any stress would just disappear.

However it's the whole concept I'm questioning. To get stress you need force being resisted.

Where is this force and even if there is any, is it of any value that makes a difference to anything?

There is no limit on bend angle from a cold bend, more practical things such a the area needed.

The thing you're testing from the bend is whether the cold working of the material has resulted in significant changes of material properties, especially uts and elongation.

Did you say anything over 20inch diameter people wanted hot bends? Sounds like they don't understand the principles.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[weldstan]

In a cold bend strain varies from tension (highest at extrados) to compression (largest at intrados). Determining residual stress anywhere in the bend is somewhat like that proposed by BigInch. In short it will be not worth the effort to calculate.

 

[chicopee]

Metals that I am familiar have strain stress curves. Once you calculate the strain within the plastic region, you can figure out the residual stress which in due time will relax, however, you still have a ball park value.

 

[dhengr]

Auzie5:
To answer your basic question...., you can’t remove a coupon from the pipe without changing the residual stress in it and in the pipe around the coupon hole. And, in fact, if you could measure accurately enough the bend angle in the pipe would change from cutting out the coupon, because you have disrupted the equilibrium stress conditions (the residual stresses) after the bending process. I am aware of the drilling and strain gage method of assessing residual stress, and it pretty much seems to be the standard method today. It machines a std. hole out of the part and measures the strain change around that hole to determine the residual stress. I’ve never used the system you linked and haven’t studied the subject in some years. We didn’t have nearly as nice a set-up as shown in the video, but what they are doing is the same process as we were studying and learning.

Try this thinking on for size, and see if it fits your needs or understanding of your problem. At the same time dig out your Engineering Mechanics, Advanced Strength of Materials and Theory of Elasticity text books and start reviewing them on these topics. Let’s also start out with a simple rectangular beam section, in simple bending, a simple max. bending moment at the center of the beam. We’ve all seen this problem in our early engineering courses, so it may be a slightly simpler example, but the thinking is the same for a cold bent pipe, just a bit more complicated because of shape geometry, etc. The beam is loaded until the top and bot. fibers just start to yield, that’s L1 (load 1). If you continue the loading the locations (the planes) of the yielding fibers move inward toward the N.A. (neutral axis), that’s L2. You are starting to form a plastic hinge, but if you unload the beam, much of its deflection (Δ) will spring back, but not all of the deflection. This residual deflection (residual curvature) is because of the fiber blocks, t&b which have gone beyond yield, have taken a permanent set, and do not (can not) return to zero. And, in fact, material immediately adjacent to these yielded blocks is affected by the block’s yielded condition, and can not return to their initial fiber length either, and the sum of these is residual stress. If you reload the beam to L2, the beam will go back to the deflection which existed before, at L2. If you continue loading now, to L3, the deflection will increase and the two yielded stress blocks will grow, moving further toward to N.A. Now, unload the beam and a greater residual Δ will remain and the residual stresses will be different too, and these stresses will be at a new equilibrium within the member.

Now, look at a typical Stress/Strain curve for most common steels (metals) that we use. They show some linear slope upward as stress and stain increase from zero, that slope is the Modulus of Elasticity (Young’s Modulus), the material follows Hooke’s law; some construction steels have a distinct ‘yield point,’ and then a long plateau with a slight upward slope as stress and strain increase, before they reach a strain hardening region on the curve, at some significant strain. Many steels and other metals don’t have a distinct yield point, so we usually define a yield strength at a .2% strain offset on the curve, and their curves have some straight upward slope from zero stress & strain and then a more continuous upward curved shape, a gradually increasing strain hardening range. You should be able to find the above in some more detail in a good text book. ASM also has some good materials on the subject for various materials. Now, if we think of our beam again, and plot our loadings on the Stress/Strain curve; L1 happens at or near (a little below?) the yield point or yield strength of the material and if we stop just short of L1, the material will unload right back down the slope “E” (the Modulus of Elasticity) slope line. Now, if we reload the beam to L1 and continue on to L2 we will move up the plateau or curve to some higher stress and strain point, and if we unload the beam now, the material will unload from the L2 point on the curve, following a slope of “E” (Hooke’s law) & (parallel to the original “E” slope, to zero load/stress, but with some greater residual strain related to the residual beam Δ or residual stresses. If you reload the beam again, the material will move up the right-shifted “E” sloped line to the L2 point on the curve and then follow the curve up to the L3 load point on the curve. This reloading can basically be repeated on up the curve until the material fails. In fact, this is what you are doing every time you re-bump your pipe with the hydraulic ram, to increase the bend angle a little more.

This yielding and the residual stresses are usually not a problem for our daily design problems because the yielding and residual stresses are nicely aligned (oriented) with the primary stress fields of our loading in the structure/pipe in use. We bend the beam or pipe and release the hydraulic ram (unloading) and the material unloads from L2 or L3 on down the “E” slope to a significantly increased (right shifted) strain. When we apply our design loads to the beam or pipe the material moves up that “E” slope toward the L3 point on the stress/strain curve, never reaching that point if our design stresses stay below yield, or then continues up the curve if our design stresses exceed yield or the L3 stresses, as we apply our design load, and we hardly know the difference. Cold bending is usually not a problem under reasonable conditions, and most of our designs produce stresses below yield, so they are operating on the lower portion of the Stress/Strain curve, with no problem. We are operating below L2 or L3 on the Stress/Strain curve, but starting from a right shifted strain at zero load, and we are still operating well below the ultimate strength of the material. Alternatively, from some situations, you can get poorly arranged (aligned) or very high residual stresses, very stiff/restrained connections, nasty triaxial stress conditions, bad welding and welding details, where this all goes to hell in a hurry.

I think, that rather than testing to find residual stresses at one location for this kind of problem, you might be better off explaining the concept of this residual stress and how it comes to be, as I’ve tried to do above and let it go at that. When we cold bend the beam or pipe we can make reasonable stress calcs. for our bending operation, and then for our unloading operation, and finally for some residual stresses under simple enough conditions. When we reload it under design conditions we don’t allow the stresses to get that high again under normal conditions, and we know it’s starting at some shifted strain level and following the normal Modulus of Elasticity slope, and we know that under normal conditions this is not particularly detrimental to our designs and end uses. To start to actually try to measure these residual stresses and report on them on a regular basis, unless you are addressing a real significant issue or problem, may be starting something that quickly gets out of hand. You’ll never get a string of pipe built for actual use, becuase customers will want you to make 10 tests at every bend, so they can try to prove that they have reason to doubt your analysis and your engineering judgement. This has been working for years with normal pipe diameters, and bend radius’. I’m all for testing if it will prove something, but not just to increase costs, when there is no problem to be evaluated, and nothing to prove.

 

[LittleInch]

dhengr,

great post.

auzie5 - perhaps you can explain a bit more what it is you're actually concerned about and why. Cold bending of pipelines of all sizes has been occurring for decades without any need to test for or take account of non existent "residual" stresses, so what has changed?

Also I'm interested to know what sort of technique you're looking at. The hydraulic ram machines I can find are quite simple small bore affairs. Given that you need to extend the pipe beyond the radius you want to be left with to take account of the yield springing I'm not sure how that works if you're just pushing a radiused die into the middle of a pipe, but Maybe I'm mistaken in your approach.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[dhengr]

LittleInch:
You said..., “Cold bending of pipelines of all sizes has been occurring for decades without any need to test for or take account of non existent "residual" stresses, so what has changed?”

I agree with you that residual stress testing should not be needed under normal conditions. It might be useful if the OP’er. has a real problem to resolve, but not for normal pipe materials, diameters, bend radius, bend angle, etc., using regular cold bending equipment and methods. But, the expression “non existent "residual" stresses” is not correct, and it has kinda been implied several times above. It is generally invisible to most of us, under normal conditions, because it causes absolutely no (no significant) problems in our final designs. But, if you are cold bending pipe or steel beams residual stress does exist in these members when you are done, that’s what causes them to keep their new/final shape. Many of these members have residual stress from their original manufacturing process too. Again, we just don’t see them or have problems with them. Rolled steel beams come to you with lots of residual stress locked in, mostly from thermal (cooling) stresses after hot rolling. Hot rolled stl. pl. has fewer, less complex, residual stresses (not, no residual stresses), cold rolled stock has more residual stresses than hot rolled, and then we start bending and shaping them to fab. them. Much like the OP’er. removing his coupon for testing, you will see residual stresses show their ugly head when you try to machine a large fabed part, because the machining releases or rearranges some of the residual stresses and you can’t hold final tolerances. Then a stress relieving heat treatment is called for before final machining. Hot bending will take care of some residual stresses, and bending problems, but it can introduce new residual stresses of its own.

Some residual stresses, in some types of materials will relax a bit with time, transport and handling vibrations and flexure, and heating/cooling cycles in operation. But, the whole existence of residual stresses, their distributions and magnitudes is kind of a fickle matter, so one size does not fit all. We do know from experience, that in many cases their existence does not cause us serious problems in the use or design of our normal structures. Other than that type of statement, I don’t go around kicking sleeping dogs, and looking for trouble or another life long thesis subject. I would try the layman’s explanation of residual stresses on a client/customer, and tell them to let me worry about when they might become an significant issue. Sometimes, they hear something or read a snippet written by a not-tech writer, and get all worked up over nothing, and I don’t want them causing twice the work or cost than what is needed.

 

[LittleInch]

Dhengr.

Again a great post and puts it in context. What I really meant was that residual stress is not a common issue unless you do something that allows it to be seen such a cutting a large hole or cutting a section of. At a free end or with free flexing, we've all seen "spring" from pipe which can require considerable force to realign it, but in stress terms is not that significant.

In a cold bend the bend stays in place because the material has yielded and plastically deformed. I'm struggling to see where residual stress really comes into this.

Even if you can measure this what is the purpose?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.