Flat Heads and UG-34

[zakk]

We are designing an assembly subject to internal pressure that will use a circular flat head that is retaining mechanically at the outer diameter (Fig UG-34(m)). The head must meet Sec VIII Div 1. Is the Equation 't = d*sqrt(CP/SE)' only applicable to the plate center where maximum bending occurs, or does this minimum required thickness have to carry all the way to the plate periphery?

 

[Trestala]

That equation, from my understanding, applies to flat heads with uniform thickness (or assumed uniform thickness, i.e. ignoring raised faces, etc.).

The resulting minimum thickness should carry along the periphery.

When you have varying thickness, the stress distribution will be different, and the analysis would be different.

 

[michaelb7]

I've used the following formulas in the past:

Mx = (p/16)[(d2/4)(1+µ)-(r2)(3+µ)]
Mz = (p/16)[(d2/4)(1+µ)-(r2)(1+3µ)]
From Process Equipment Design: Vessel Design, p102 eq 6.50 and 6.51

σx = 6Mx/t2
σz = 6Mz/t2
From Process Equipment Design: Vessel Design, p102 eq 6.55

Plug in the allowable stress for σx, σz. Solve then stress equations for t. Then combined with the moment equations to get the required thickness of the cap as a function of r. Use whichever t is larger, for any given radius.

I'm not entirely sure if this is legal for the ASME Boiler and Pressure Vessel Code. I'm rather new to engineering, and I don't have a good mentor at my workplace, but this method seemed to make sense to me.

 

[SnTMan]

zakk, what is it you are wanting to do?

If you want to 1) vary thickness w/ radius, I would say UG-34 does not support this, although it is true greatest thickness would be required in the center. As well, it seems un-efficient to manufacture such a head as opposed to cutting a circle from plate.

2) Thin at the edge, such as for a groove, if so see UG-34 under Sketches (j) and (k).

Regards,

Mike

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

 

[zakk]

SnTMan;

What we are trying to do is design a clamp connection similar to Appendix 24, except replace one of the Hubs by a Flat Head. I attached a sketch of a general configuration. According to Fig UG-34(m) for a mechanically locked Head, the diameter "d" should be taken as the outside diameter of the Head (Dimension "C" in the sketch). I take that to mean t_min should taken all the way to the periphery of the plate. I have seen numerous ASME Code parts, namely flat head closures, where this is not the case. The Head is held in place with a lip inserted into a clamp ring and the lip is a fraction of t_min.

I discussed this with my AIA. They agree that the maximum bending stress is located at the plate center, however the stress (and required thickness) as one moves away from the center is a grey area and if I can rationalize the thickness reduction, they will consider it. The question is... how? I could break the Head into a series if concentric rings and calculate the moments and shear at each interface to develop a system of equations (Roark & Young, 8th Ed Para 11.7). If someone can walk me through that I'd appreciate it.

We are considering burst testing (UG-101(m)) the assembly to address any ambiguity in the design. However, since I can't use burst testing to override Code rules, I'm back to my first question... is the intent of UG-34(m) to carry t_min to the outer periphery of the plate?

 

[SnTMan]

zakk, I am not at all familar w/ Apx 24 and these connections, however:

If designed per UG-34 I'd say the "d" dimension is equal to the pressure seal diameter, dim "A" in your sketch, as per Fig UG-34, sketches j & k. I'd say you need to maintain tmin to this diameter. Then perhaps the thickness of the section from A to B would be subject to the thickness under the groove per these sketches. I would take the lip thickness from B to C to be subject to Apx 24 rules.

Alternatively, you could decide the whole thing is a U-2(g) design. Roarke, I believe, has equations for thickness of circular plates as a function of radius, which you could presuambly use. Don't know how or if Apx 24 would relate to this approach.

As to the concentric rings approach I guess you'd equate the deflections and slopes either side of the interface and solve the systems of equations for thickness. I guess. Selecting the proper edge condition could be troublesome. Again, don't know how / if Apx 24 would relate.

Not much help, I'm afraid

Mike



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