ASME VIII-1 Appendix 2 integral flange (Figure 2-4, Sketch 5/6/7) - hub internal to shell boundary permissible? 1

[COpressurepipe]

I have a unique situation for a large diameter waterworks control valve installed within a vault. The valve flanges are raised face w/ a bore much smaller than the nominal diameter of the adjacent pipe cylinder. The pipe cylinder OD is fixed & cannot be changed without affecting hydraulic performance of the system. Use of a loose ring type design (Sketch 3/3(a)) isn't desirable due to the low gasket contact width (~1/4"). Per appendix 2 is a design placing the hub on the 'inside' to the attached cylindrical shell valid? - attached is a diagram c/w the resultant locations. If yes, must the hub be designed such that g_o=g_1 as shown in Sketch 5?

Although the sketches under Figure 2-4 show the hub on the 'outside' in all instances it isn't explicitly stated anywhere in appendix 2 that the hub must be located externally... or am I mistaken here? Deign pressure is 550 psig; flange & shell materials are SA-350 LF2 & SA-516 70, respectively.

 

[r6155]

No problem.
Regards

 

[Christine74]

I think you're going to have problems here. As you stated nothing in Appendix 2 (or any code for that matter) provides guidance on how to design a hubbed integral flange with the taper on the ID, so you would have to approximate it by setting g1 equal to g0. But the problem you'll have is that your taper angle would be 0 degrees in your case and the calculated Appendix 2 stresses are highly dependent on the hub taper angle.

I think you'll find that you need to include a taper angle on the OD of at least 3 or 4 degrees in order to end up with a flange design that's somewhat reasonable.


-Christine

 

[Geoff13]

In your sketch there is a note pointing to the taper face that says "epoxy". I'm going to assume that the steel portion only comes to the ID of the attached pipe, and the entire taper is epoxy. If this is the wrong assumption then my comment will also be wrong. However if it's correct, I would ignore the epoxy zone in terms of the Appendix 2 calculations. I don't know if this leaves you with any gasket seating width. However, I would expect that having the forging manufacture include the taper area when they machine out the shape would be cheaper than all the work to add this epoxy taper later.

I would need to think carefully about when to use the taper ID (B in your sketch) vs. when to use pipe ID. Simply using the taper ID as B in all the Appendix 2 formulas will give wrong answers. You must think about what force or moment arm each formula is trying to calculate and select what's appropriate for your non-standard geometry. I don't know if reviewing 2-13 will give any better understanding of how to use appropriate values.

If the taper is epoxy I would likely treat this as Sketch 7 or 11 (integral), without changing the pipe butt weld detail in your sketch. Once welded these are all the same. Sketch 7 allows g0 = g1 with a radiused corner.

No matter how carefully I thought about this I would be reluctant to end up with a geometry that has stresses very close to the allowable. If this flange saw 30 psi (gentle leak) instead of 550 psi I might be a bit more liberal in how close I would get (This is the "can I sleep at night" factor). It is impossible to know what geometry assumptions are hidden in the F, T, U, V, Y and Z factors.

This is an interesting problem. Good luck.

 

[LittleInch]

What is the mating face on the valve look like or are dims C, X and B the same? i.e. a large raised face area?

What was there before? Or is new?

Certainly interesting.

 

[COpressurepipe]

In your sketch there is a note pointing to the taper face that says "epoxy". I'm going to assume that the steel portion only comes to the ID of the attached pipe, and the entire taper is epoxy. If this is the wrong assumption then my comment will also be wrong. However if it's correct, I would ignore the epoxy zone in terms of the Appendix 2 calculations. I don't know if this leaves you with any gasket seating width. However, I would expect that having the forging manufacture include the taper area when they machine out the shape would be cheaper than all the work to add this epoxy taper later.

I would need to think carefully about when to use the taper ID (B in your sketch) vs. when to use pipe ID. Simply using the taper ID as B in all the Appendix 2 formulas will give wrong answers. You must think about what force or moment arm each formula is trying to calculate and select what's appropriate for your non-standard geometry. I don't know if reviewing 2-13 will give any better understanding of how to use appropriate values.

If the taper is epoxy I would likely treat this as Sketch 7 or 11 (integral), without changing the pipe butt weld detail in your sketch. Once welded these are all the same. Sketch 7 allows g0 = g1 with a radiused corner.

No matter how carefully I thought about this I would be reluctant to end up with a geometry that has stresses very close to the allowable. If this flange saw 30 psi (gentle leak) instead of 550 psi I might be a bit more liberal in how close I would get (This is the "can I sleep at night" factor). It is impossible to know what geometry assumptions are hidden in the F, T, U, V, Y and Z factors.

This is an interesting problem. Good luck.

'Epoxy" just denotes the anti-corrosion coating to be applied after welding to the shell. Illustration was to show the transition to the cement mortar lining of the pipe cylinder. The entire flange will be a SA-350 LF2, CL1 seamless forging.

I agree that the solution seems to be developing forces (H_D, H_G, H_T) onto an exploded free-body diagram to verify the moment arms. Also agree it seems prudent to use the shell ID basis for hydrostatic end force rather than the bore ID.

Your observation that there's inherent 'mystery' to the hub factors did cross my mind too... would be helpful if anyone could point to the historical basis for these. Likely a bulletin or journal entry from years ~1900-1925...

 

[COpressurepipe]

What is the mating face on the valve look like or are dims C, X and B the same? i.e. a large raised face area?

What was there before? Or is new?

Certainly interesting.

Mating face shown matches valve verbatim. Gasket material to be Garlock Blue-Gard Style 3000 (m=5.2; y=4,400 psi) @ 1/8" thick. Final bolt stress during assembly
is anticipated to be very close to allowable per ASME II-D. In fact the design might need to allow for fully developed bolt load, (Ab*Sa), rather than the 'average area' method in equation 2-5 (3.e.5).

This is a new installation. Diameter is of a magnitude that use of an ASME B16.47 CL300 Series A RF weld neck is not possible.

Very atypical situation, yes...

 

[Christine74]

"Your observation that there's inherent 'mystery' to the hub factors did cross my mind too... would be helpful if anyone could point to the historical basis for these. Likely a bulletin or journal entry from years ~1900-1925..."

Close, the procedure for designing flanges was outlined in "Formulas for Stresses in Bolted Flanged Connections" (Waters, E. O., Wesstrom, D. B., Rossheim, D. B., and Williams, F. S. G.) in the 1937 edition of Transactions of the American Society of Mechanical Engineers.


-Christine

 

[LittleInch]

The gasket data means nothing to me I'm afraid, but the question I have is whether you really need all that gasket bearing area? Other than maybe a little bit of flow disturbance which you could fill in the gap with something non structural or just machine off some of the inner RF, is there an issue with using a higher gasket pressure and making the RF bit of the flange much more like a "normal" full size flange?

At the moment you have a 4.5" RF gasket bearing distance. Could you make that say 1" with the same X distance?

Then it makes it much more like a standard flange assessment.

 

[Geoff13]

Close, the procedure for designing flanges was outlined in "Formulas for Stresses in Bolted Flanged Connections" (Waters, E. O., Wesstrom, D. B., Rossheim, D. B., and Williams, F. S. G.) in the 1937 edition of Transactions of the American Society of Mechanical Engineers.

Wow. 40 years doing this, and I never knew the source. Thanks.

I note that almost 90 years later ASME still wants to sell a copy. Apparently they feel it never falls into the public domain. It might be an interesting read, but not interesting enough that it's worth spending money.

 

[COpressurepipe]

Thank you for providing the reference.

I think shrinking the gasket area is a sound direction to explore...should be a lower installation load for the bolting. Bore could be as small as 72.0" & utilize a g_1=1.5".

Got to thinking about the moment configuration on an integral reverse flange shown in 2-13 as mentioned above... Is a reverse-style flange defined by the shell pressure boundary (dim. 'A' Figure 2-13.1) lying outside the bolt circle, or rather dim. 'A' lying outside of the gasket reaction (dim. 'G' Figure 2-13.1) location? Code only describes as the "configuration indicated in Figure 2-13.1"...

Two observations I think may need to be considered in the design:
- when the hydrostatic end load for H_D utilizes dim. 'A' instead of dim. 'B', H_T acts in the opposite direction & should be taken as zero
-'h_o' parameter for hub taper geometry is a function of dim. 'A' rather than dim 'B' for integral reverse flange

 

[r6155]

Again: NO PROBLEM

 

[COpressurepipe]

Again: NO PROBLEM

After further study I'm inclined to agree... maybe you could help me with one last issue I'm trying to eliminate.

For the 'h_or' factor - since the inside radius of the shell is only slightly larger than dimension ' B' ', is it not reasonable to assume that dimension 'B' (or even the shell OD) is to be used rather than 'A' per Figure 2-13.1 as the shell does not extend to the flange OD? Stress outputs tend to indicate the difference is pretty small comparing 'B' to 'A' when calculating h_or.

H_T is most definitely positive, and h_T should fall between h_D & h_G. I believe the 'aplha_r' multiplier addresses the reverse direction of the hub taper via factors T_r, U_r & Y_r, yes? The reverse factor values are very close (order of magnitude) using 'A' when calculating h_or, compared to 'B' when calculating h_o.

This 'flange influence factor' - h_or, appears in similar fashion for influence width of anchor rings attached to thin shells. Inside radius of shell is used in this application.