Cylindrical Tanks - Out of roundness 1

[AnthonyMarx]

I am seeing this phrase in specifications for tanks and other vessels.
"The difference between the maximum and minimum inside radius, measured from centre of the tank, shall not exceed 0.5% or 37 mm whichever is less"
Can anyone let me know the code this originates from.

 

[GregLocock]

Given the spelling of centre I'm guessing UK. Quite how you establish the centre of a non uniform roughly circular thing is more a of a debate than a process.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[IFRs]

Data clouds from 3D laser scanners can be resolved into curves that represent the tank shell at any elevation. Mathematically constructing a best fit circle from this curve gets you a tank center. Evaluating roundness is now straightforward for that elevation.

The same can be done for the entire tank - a best fit cylinder can be created from the data from which lean, deviation and roundness can be derived.

There are many ways to slice the data - another is to create a best fit curve, circle and cylinder from the lowest 12" of the tank, and evaluate lean, deviation and roundness from those.

For example, API 650 measures roundness at 3 feet above the bottom and has roundness tolerances that vary with tank diameter, allowing greater variations in larger tanks. For this example, roundness from a data cloud as described above is a fairly straightforward exercise.

The OP's question was more of "where did this tolerance come from?" My guess is that it is a way to insure quality construction and suitability for service rather than some structural requirement or FEA analysis. Sometimes an organization just picks a number out of thin air and over time either adjusts or sticks with it noticing that it "works", and thinks no more about it.

 

[dvd]

I have no idea the possible source. It seems like a good way to spend a few hours arguing, though.

Who gets to decide where the center is located?
D/2 - d/2 = 37 mm - ok
D/2 - d/2 = 0.5% - not ok, 0.5% of what?

 

[JStephen]

"Pressure Vessel Dimension Inspection - Out of Roundness of Shell: UG-80 deals with out of roundness of shell. The maximum permitted ovality tolerance (D max – D min) shall not exceed 1% of the nominal diameter of the vessel."
That's from an online discussion of the ASME code, which I don't have in front of me, but that's basically a 1% on diameter split into maximum out and maximum in, which matches your 1/2% on radius to an extent.
I recall reading long ago (maybe Martin Gardner books?) about them getting into trouble measuring max OD vs min OD on submarine hulls, not realizing that there were constant-width shapes that were also not round.
I would think any actual tolerance would be based in large part on what is achievable with regular production methods and good workmanship.

 

[TGS4]

I am not so sure about tanks, but pressure vessels have very specific requirements regarding out-of-roundness. And there are different values depending on whether the failure mode is plastic collapse vs buckling. For internal pressure, the out-of-roundness can add additional stresses, that may exceed allowable values. For external pressure, the out-of-roundness may lead to premature buckling.

BTW, the values are typically measured as 1% difference in diameter measurements.

If you have a more specific reference, please let us know.

 

[GregLocock]

IFRs, exactly. The local radius of curvature of a line is not the radius as measured from the centre of the section, partly because of the difficulty of defining a centre, even if you specify in addition that curvature must always be positive for the entire perimeter, ie no re-entrants. The OP's definition raises mathematical problems far beyond its intent, I suggest, that is, it is poorly worded. It could be made workable by defining how the centre is established, perhaps it does.


Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[JStephen]

I would suspect it is worded the way it is specifically to avoid the constant-width-but-not-round issue.

 

[Denial]

I once had the problem of establishing a radius (and a centre location) for an arc that was approximately circular.[ ] The problem arose when designing an upwards extension to an old arch dam.[ ] Surveyors had determined the exact location a series of points along the mid-width of the top of the existing dam.[ ] It was, conceptually, a simple least-squares exercise in three unknowns, these being the X & Y locations of the required centre-point and the radius.[ ] The "error" being minimised was the sum of the squared distances from each surveyed point to the circumference of the assumed circle.[ ] These days it would be a ten minute exercise using Excel and its Solver.[ ] Back in those days it required the tedious development of a Fortran program.

 

[GregLocock]

My particular thought is this charmer

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Andrew ONeill]

It will be historical. Keep in mind most codes were created in the 1900's, well before computers and widespread survey equipment.
To measure out a tank base, a peg would be placed in the middle, and a string used to drawing the diameter.
37mm just happens to be almost 1.5 inches, so that's likely where the number came from. Likewise the 0.5% of radius, is likely an arbitrary value based on judgement and experience.

Just my 2 cents.



Andrew O'Neill
Specialist Mechanical Engineer
Rio Tinto
Australia