How to draw 2:1 Elliptical Head 3

[ColinPearson]

Hello folks. I want to see if anyone has a good reference for drawing a 2:1 elliptical head. I always draw them in CAD and get the volume to calculate the weight, but it's kind of a pain. I've got two xeroxs out of old layout books and they offer slightly differently rounded versions of the same approximation. Sometimes that's close enough but other times I'm left having trouble b/c my polylines won't join to make a revolution solid, or I can't make a region out of the shape, etc. That's AutoCAD specific problems as I don't have Solidworks and I still suck at Inventor.

So, good folks of the internet, does anyone have any older style ways to draw these suckers i.e. laying the thing out with arcs and bisecting etc? Or do I stick with my approximations that I've got. I'm not interested in gaining a potential tiny tiny tiny bit better approximation, I just want to find a way that makes the drawing easy.

THANKS!

 

[SnTMan]

Have you tried constructing per UG-32(d)?

You can just draw 2:1 ellipse in ACAD, but yeah, can't join the plines.

Regards

Mike


The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[ColinPearson]

SnTMan, I think that's where the approximations I have come from, and it's usually not too bad, just wondered if there was a 'traditional' drafting method that people used. Thanks for the reply.

FWIW, you can set the system variable PELLIPSE = 1 and get a polyline approximation instead of a mathematical ellipse in AutoCAD. Or, if you set PELLIPSE = 0 and draw a mathematical ellipse, you can offset it some amount and it degrades to a spline which you can then polyedit into a polyline. Not sure if you knew that, but I hope it helps someone.

 

[JStephen]

If I remember right, ASME says a 2:1 head can be constructed as a torospherical head with crown radius of 0.9D and knuckle radius of 0.17D. That's the way I draw them. Calculate inside crown radius, knuckle radius, flange radius. Draw a circle with crown radius, draw a line through the center, offset it to each side for the flange radius. Offset both those lines and the crown radius by the knuckle radius and that intersection is the center of the knuckle arc. Draw it in, and start trimming.

I figure since a head manufacturer can make a head either "perfectly" elliptical or this shape, this is just as likely to be "right" as a true elliptical shape and simpler to draw.

The problem is that if you offset an ellipse, you don't get a bigger ellipse, and that messes things up.

 

[r6155]

"An acceptable approximation of a 2:1 ellipsoidal head is one with a knuckle radius of 0.17D and a spherical radius
of 0.90D." UG-32
This is only for drawings, not for design, manufacturing and inspections.

Regards
r6155

 

[ColinPearson]

Thanks for the input folks. Looks like I've pretty much been doing the same thing others do. Y'all have a good day!

 

[TGS4]

This is only for drawings, not for design, manufacturing and inspections.

That's an interesting claim. Please explain.

 

[marty007]

If you get tired of redrawing it every time, you can just copy/paste/scale from other drawings. Just keep a standard block for a 2:1 head and scale as required. Both the radii are related to the ID, so scaling will be accurate (except for straight flange, but that's easy to fix).

 

[SnTMan]

r6155, yes, please cite your reference.

Regards,

Mike

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[r6155]

ASME VIII Div. 1 My last ed. 2011
My example: DP= 1000 psi, ID=60 in L= 0,9ID= 54 in, r=0,17ID= 10,2 in, S= 20000 psi SA 516-70; E=1

Ellip 2:1 tmín= 1,43 in PWHT: not required, only preheat

Torispherical tmín= 1,8 in PWHT: Yes

Hence, ellip 2:1 head is not equivalent to a torispherical with L= 0,9ID and r=0,17ID
They are equivalent only for drawings.

Regards
r6155

 

[SnTMan]

r6155, no need to drag torispherical heads into the discussion. If an elliptical head is approximated per UG-32(d) in it's actual fabrication it meets Code and is acceptable.

Regards,

Mike

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[TGS4]

SnTMan and r6155, interesting that this issue should be discussed at this specific time. At the ASME Boiler Code Week, next week, I have an item to discuss exactly this approximation provided in UG-32(c).

[As an aside, in the editions and addenda prior to 2015, this paragraph was sub-paragraph (d). However, in the 2011 Edition paragraph UG-32(b) was deleted, and the UG-32 subparagraphs were subsequently re-numbered for the 2015 Edition. If you look at the 2015 Edition, you will see that it is UG-32(c) now.]

As r6155 rightly points out, the calculated head thickness for the "approximate" torospherical head is higher than the "actual" 2:1 semi-elliptical head. So, the point that I will be bringing up is whether or not that approximation is appropriate. Before I do my presentation there, does anyone here have any thoughts or comments?

 

[r6155]

I insist. UG-32 in not used as tolerance. ¿Which is the meaning of "approximate" here?. The correct term is "out-of-true", hence UG-81 "tolerance of formed head" must be applied.

Regards
r6155

 

[ColinPearson]

Hey Marty007, I never really thought about it but I guess the dimensions are all linearly related to teh ID as you say, so I will definitely keep head lying about to copy/scale... thanks!

 

[pdiculous963]

TGS4-In regards to the torispherical approximation, it is worth noting that this is how Division 2 handles the design of ellipsoidal heads. The required thickness for r6155's head would be about 1.37 inches using these rules with the Div. 1 allowable stress at 70°F. While they may be considered the same from a code calculation perspective, I'm sure the collapse pressures predicted by a FEA would be different.

 

[abuanaselmasry]

Code approximation of crown 0.9D & knuckle 0.17D doesn't result exactly in 1/4D depth of the head which is basic dimension as the name of the head is 2:1. The reason for this is that the 0.17 has some other digits. So, I'm using SolidWorks to draw the head. I set the crown dimension and the depth, then knuck radius will be decided automatically by the geometry. It is normally about 0.171875D.

I can help you by giving you a sketch (*dwg format) generated by Solidworks in a dimeter, let's say, 1000 mm. Then you can scale it later on for any diameter will use.

Regards,
M.Salaheldin
Static Equipment Mechanical Design Engineer

 

[JStephen]

I have assumed the intent of the approximation statement is that heads can be produced to approximately either shape and be considered equivalent.
Whether the thickness calculates the same applying the torospherical equation to the approximated elliptical head is a moot point. A similar discrepancy occurs when applying flange rating equations to 150# flanges, I believe.

TGS4, I think a major consideration would be to check how head producers actually produce heads. If they all produce "true" 2:1 form or all produce 90/17 or whether some do it one way and some the other.

 

[r6155]

Ellipse has infinite radius. When check tolerances with UG-81 it is impossible to measure infinite radius, hence an approximation 0,9D / 0,17D is used. Not exist spherical portion of ellipsoidal head.
If purchase order state ellip 2:1 head the manufacturer shall produce ellip 2:1 , not 0,9D / 0,17D(torispherical)

Regards
r6155

 

[JStephen]

I went ahead and submitted a request for Interpretation on that point. #15-2512.

 

[TGS4]

Thanks JStephen - you can also follow my item as 15-18341104 2177. If you don't object, I am going to request that your item be covered under my item, as well.