Checking bend angle and prep trueness

[Supe]

We have an issue with our manufacturing facility, where we're not getting induction bends that are exactly the 90 or 45 degree angles specified. This is compounded by the fact that the weld preps they're cutting are not always true to the pipe, i.e, they are angled on some sort of plane.

What I'm trying to figure out are some ways of checking these induction bends for dimensional accuracy. Diameters are as large as 32", with wall thicknesses as heavy as 5". Even if the bends are off slightly on the pipe, what is most important to us is that the weld preps are accurate to the specified angles in all planes. Basically, we need to make sure that the weld preps are true so that we won't have a big root gap tolerance when we align the bends with the lengths of pipe they are to be welded to.

I'd love to hear of any good ways to check for this, whether it be some sort of adjustable jig/fixture, laser lines, etc. Please keep in mind, there will be no way to level the components themselves, and the check must be able to be performed by a single person.

[A_DAB_will_do]

Foundation questions:

1. What kind of money are you willing to spend to solve the problem? ?seriously is this a $100, $1K, or $1M dollar problem?
2. How fast do you need answers?
3. What's the tolerance for the part dimensions and acceptable error in the measurement for your purposes?

I can think of a couple solutions that are viable, if you have $$$ to spend. There's some fantastic equipment available for optical imaging and 3D modeling that could(once it's set up) give you fast, accurate evaluations of groove geometry, pipe shape distortion, and many other dimensional characteristics.

But, the solutions I'm familar with may be moot if they're too costly, slow, or innaccurate for your needs...

[Supe]

Needs to be low buck, could probably spend a few hundred dollars on materials if need be. Way ahead of you imaging software/equipment, but I need something that can be implemented very quickly and can be either fabricated cheap enough to produce multiples in at least two manufacturing facilities or job sites, or small/light enough to be shipped or carried from the east coast to Colorado on a whim. The closer we can get to a go/no go type solution, the better. If its a solution that creates some sort of visual indicator other than a measurement, that'd be ideal. Over at corporate HQ, visuals speak volumes more than numbers on a print do.

[A_DAB_will_do]

sounds like you're in a tough spot. cheap, fast, and good....not too many solutions to problems that work for all 3...

Re-reading your original post, it sounds like you have 4 problems, not one. Feel free to correct me if I'm wrong here.

1. Inaccurate bends when fabricating 45° and 90° bends using induction heating.
2. Bevels on pipe that are not square to the pipe axis in one or more planes.
3. How to measure #1
4. How to measure #2

Pipe diameter and wall thickness varies.
Pipe sections cannot be moved to minimize hi-low, root opening variations, etc.

Fixtures that are go-no/go gauges are probably not practical since you have lots of different size parts to check, and you need the solution to be portable. Multiple gauges so you don't have to move them around are a problem since you could spend lots of time ensuring that they give identical results when testing parts.

I'm thinking the simplest way to check the bends is to measure the outer circumference at a prescribed height or part midline. Your parts are large enough that using a tape measure is probably accurate enough. Even hardware store tapes are pretty accurate these days... Someone could calculate what the circumference is for 90° and 45° +/- 1°(or whatever tolerance is needed) for all part sizes. This gives you a table of values that can be checked in the field, as a go/no-go gauge before starting work.

Checking the straight pipe bevel alignment is tough if you can't level the pipe to start off. All the tricks I know of start with leveling the pipe and then making measurements from the 12,3,6, or 9 o'clock location of the pipe OD. maybe someone else has a trick they know.

At the risk telling you something you may already know, it seems to me like to might be better off getting back to the root problems. Fix the process that makes the 45 and 90 bends. And fix the process that produces the bevels on the straight pipe sections. If these are right, then there's little or no need to measure these things in the field...

You may need to make the measurements to fix the root problems, but then again maybe not. It's just my opinion but I think your time, effort, and creativity are better spent on fixing the root cause of your fit-up problems, rather than figuring out the perfect way to measure how bad the problem is in the field...if I'm preaching at the choir, please forgive me.

Hopefully someone else here can give you an easy way to measure problem #2 within your constraints, and save you from re-inventing the wheel.

-Dave

[Supe]

My thoughts are right alongside those of yours. Unfortunately, fixing the root problem isn't really an option without first reinventing the wheel.

Unfortunately, politics are the limiting factor here. Company CEO started out with the fab shop business, which over time turned into a Fortune 500 engineering and construction firm. QC KNOWS that the problems are fab shop problems, but because the fab shop is the CEO's "baby", it's either a "find a way to document and prove it," or "shut up and deal with it in the field" type situation. They will certainly not be bending over backwards to help us prove that they are the ones screwing this up by not meeting spec.

I think what's it going to come down to is making a protractor-type template for measuring bend angle so that any visible gaps can be seen, and just using a large square at various positions along the weld prep from the pipe OD to check that it is perpendicular in all planes. Relating the two together will likely be out of the question. Measuring bend radius with a tape is not an option, as not all elbows have equal length legs.

[A_DAB_will_do]

I understand where you're coming from and have been in your (unenviable) position myself a time or two.

I discarded the notion of a giant protractor template when you mentioned how large the pipe is. The pipe diameters will make such a device unwieldy, to say the least. But if that the best all around solution, you gotta do what you gotta do...

I did not consider that the bends might not fall in the middle of the pipe section. I agree that this rules out using a tape to simply measure the bend circumference.

you might be able to at least document the bend variations for future analysis; evidence for proving the root cause of the problem. An overhead photo, from an exact distance could be used to take some pretty accurate and precise measurements. If the focal length is a constant, it's pretty easy to relate pixels to distance in a photo. It's even easier if you photograph a scale along with the object you want to measure.

There's some reasonably inexpensive software for making just these sorts of measurements. Check out packages like PAX-IT! or Image-Pro and you'll see what I'm talking about.

Originally Posted by Supe

My thoughts are right alongside those of yours. Unfortunately, fixing the root problem isn't really an option without first reinventing the wheel.

Unfortunately, politics are the limiting factor here. Company CEO started out with the fab shop business, which over time turned into a Fortune 500 engineering and construction firm. QC KNOWS that the problems are fab shop problems, but because the fab shop is the CEO's "baby", it's either a "find a way to document and prove it," or "shut up and deal with it in the field" type situation. They will certainly not be bending over backwards to help us prove that they are the ones screwing this up by not meeting spec.

I think what's it going to come down to is making a protractor-type template for measuring bend angle so that any visible gaps can be seen, and just using a large square at various positions along the weld prep from the pipe OD to check that it is perpendicular in all planes. Relating the two together will likely be out of the question. Measuring bend radius with a tape is not an option, as not all elbows have equal length legs.

[Supe]

Appreciate the advice, I will look into those software packages.

[gaustin]

Are the elbows able to be moved or layed down flat ? Is laying them flat on a level surface , verify that both ends are plumb, transferring a line from both faces down to the floor then extending those lines until they intersect to verify a true 90 degrees.

After verifying that the faces are correct then the bevel angle can be checked by running a straightedge or string across the face and measuring that off the string/strainghtedge.

If the faces are not at 90 degrees to each other with the ell laid flat then get one face plumb and rotate the ell until the other is plumb. Then using a water level or laser, see what the difference in elevation is .

I have never worked with a 32" 5" wall elbow but regardless of the size, you have to start by getting some surface plumb/level. I have verified square on 20' plus round fabricated elbows and imagine the same principals work.

My explaining may be off though.

[Supe]

I understand what you're getting at gaustin. The elbows will be laid flat, but the odds of getting them on a level surface is slim to none, which puts me right back where I started.

[gaustin]

The ends and heel of the bend could be shimmed as needed to get one face plumb. With that face plumb, you then shim the other face to get it plumb. You can then check that both are on the same elevation.

The other option is to place on face straight down on the floor and level that face. Then drop a plumb bob to verify the other face is plumb. That lets you know you have a true 90. With a 45 you woul have to establish 1/4 marks.

Was this machined after it was bent ?

Check you PM's

[gaustin]

Maybe this. Establish 1/4 marks at given distances from the end of the bevel . Measure the chord at the two opposite locations and compare . If these are the same true in that plane. From the other two quarters, pull wire and check for 90.

The hard part is establishing the starting point for the 1/4" marks.

I dunno. Again I have not worked on ells that big AND thick.

[William McCormick Jr]

Originally Posted by Supe

Appreciate the advice, I will look into those software packages.

This will probably not be applicable, however I had doubts about it myself for my application. We had a job to cut down the pipes that the treads are mounted to, treads that mount to a spiral staircase. Because of a misunderstanding in the contract about number of treads on a wooden staircase, that was changed to a metal spiral staircase.

The customer thought that when the stair manufacturer changed the number of treads on the spiral staircase to a lesser number, that he was getting cheated. The wooden stair manufacturer explained that the customer might have to duck to get under the ceiling height if he used so many treads. The customer kept saying that I paid for this many treads I want this many treads.

The stair contractor already ordered the spiral staircase, with each tread rise set for 13 treads. He brought them over and asked if there was any way to just skim off a bit on each tread. Some of them were going to have to be cut very close to the tread.

The only thing I could think of that would not totally screw up the existing treads was to plasma cut them. Or mount each tread on a smaller pipe with the threads for the setscrew, and hold the smaller pipe in the band saw vise. But that sounded like a hassle, and I would have had to find a three and a half inch O.D. Tubing. And waste almost half of it. The guy needed it the next day.

So I cut a 1 3/4" long ring out of a four inch inside diameter pipe, and put a quarter 20 thread into it. Got a hex head bolt and when I tightened the bolt onto the pipe I wanted to cut. I found that it was very self centering. It just totally seated perfectly. I used that as a guide for the plasma cutter.

Maybe you could make yourself a ring too. And use it as some part or portion of a device to check for square, or angle?



Sincerely,


William McCormick