ASCE "Anchorage Design for Petrochemical Facilities" ?

[TomBarsh]

I am interested in analysis of anchor bolt tensile loads for vertical pressure vessels. ASCE has a publication "Anchorage Design for Petrochemical Facilities". The initial edition was published in 1997 (combined as "Wind Loads and Anchor Bolt Design for Petrochemical Facilities", two books in one: wind design, then anchor bolt design). The new, second edition was published in 2013 as a highly-expanded document "Anchorage Design for Petrochemical Facilities".

The original edition discusses the case of vessels supported on skirts with a circular pattern of anchor bolts (para. 3.5.1.1). It provides the usual formula for bolt tension: T = 4M / (N*BC) - W/N (3.1)

It describes the conservative assumptions built into this analysis. Basically, the neutral axis of the resisting forces of compression on the concrete foundation and tension on the anchor bolts shifts away from the geometric centerline of the vessel, due to the differing material stiffness of the steel and concrete. It goes on to describe that this effect can be considered in the design (sizing) of the anchor bolts, and "may...be worth undertaking when the above equation yields large anchor bolt sizes and embedments."

The second edition of the "report" (as ASCE calls them) eliminates any reference to the "advanced" analysis of tensile load acting on the anchor bolts by considering the differing material stiffnesses. (For that matter, they have relegated formula (3.1) to an example, Example 2 Step 4).

Any ideas as to "why" they have eliminated reference to the advanced analysis? The whole report seems to be more focused on what happens below the concrete's surface, rather than what happens above. Is this why?


[The "advanced" analysis is that discussed in numerous pressure design texts, such as "Moss", "Bednar", "Megyesy", "Brownell & Young", etc, so it has certainly been accepted in industry for a very long time.)

 

[Leftwow]

bump

Are you referring to a cyclic load on the anchors?

 

[TomBarsh]

No. I don't think cyclic loads are relevant.

But it should be pointed out that this effect (shift in neutral axis) occurs only when there is a bending moment applied to the pressure vessel.

 

[Leftwow]

There was some reference in the pip code about having a "slender" vessel and how the wind affects the anchors during this situation. However, I never thought about the shift in the neutral axis. I always assumed to tension was applied from the moment centered at the edge of the baseplate.

 

[JStephen]

I'll have to look into this some time.
The impression I had was that it was based on treating the bolts and concrete around them as a composite beam, whereas the normal analysis treated the bolts only as an equivalent beam. But neither assumption is necessarily better than the other, so it was always questionable in my mind whether you were really getting a more accurate number or just cooking the numbers to give you a more favorable result.

I would be curious as to whether anyone has actually tested bolt loads in an actual vessel or tank or stack to see how they compared to either theory? It wouldn't be hard to do, just a matter of spending time and money to do it.

In the tank standards, they have added allowances for using higher-strength anchor bolts, and meanwhile, Appendix D has been added to ACI 318, and it's possible those changes just negated the need for that analysis rather than rendering it void. (The higher strength bolts are also pre-tensioned.)

 

[STR04]

Hi Tom - does the newer version support "Pretensioning?" If so, is it a little more indepth compared to the 1997 version? Thanks!

 

[JStephen]

I had a quick look at this on the way out the door this afternoon. Looks like it's not in Bednar at all. It is in Megyesy. In there, if I understand it correctly, he takes the concrete area under the base ring as the compression area and the bolt area as the tension area, and treats that as a composite steel/concrete beam. The catch is, I can't see any rational reason to do it that way. You just have that one interface where the actual cross section is more or less like that. 13" higher, you have a cylindrical shell, assume stresses there are Mc/I, so I think you'd be hard put to show stress distribution changed significantly between the tops of the chairs and the bottoms of the chairs.

I can see how you might treat the concrete as an elastic solid, consider elasticity of the bolts above that, etc., and come up with a more accurate distribution of bolt force, but that isn't what's done in Megyesy.

 

[TomBarsh]

STR04, yes, the new (current) Edition addresses pretensioning of the bolts (they call it "tensioning"). Chapter 3.8.

 

[SAIL3]

I agree with JStephan and question the logic in trying to perform a less conservative approach(if it actually really is accurate?) unless it is an existing condition and one is trying to justify the existing design ....

 

[TomBarsh]

STR04, I meant to add that the book doesn't provide detailed procedures. Also, I have been privy to several thousand vessel designs and I have seen preload applied to the anchor bolts only a handful of times. It would take special equipment and crew to install this, IMO.

 

[jayrod12]

Tom,

How specific are they expecting the pre-load/pre-tensioning to be? I've only ever seen them calibrate a run of the mill torque wrench and use that. It gets close enough for anything I need.

 

[TomBarsh]

jayrod, the ASCE text does not get into that level detail. The "Bednar" pressure vessel text discusses pretensioning of anchor bolts as a method of reducing the stress variation of the bolts under wind load, etc, in order to minimize the effect of cyclic loading. He mentions loading up to 75% of proof load.

 

[TomBarsh]

The "shifted neutral axis" analysis that I mentioned in the original post is presented in at least 6 different texts, going back to the 1960's. So I have to believe that the method has been properly vetted over time. I have gone through the derivation of it myself and reviewed a number of times and find it reasonable.

It is described (or not, in one case) in the following texts:
[ul]
[li]Moss – “Pressure Vessel Design Manual” (third edition) – Procedure 3-14[/li]

[li]Bednar – “Pressure Vessel Design Handbook” (second edition) – page 98 “Initial Tension in Bolts Neglected”[/li]

[li]Megyesy – “Pressure Vessel Handbook” (eleventh edition) – page 80[/li]

[li]Brownell & Young – “Process Equipment Design” (1959) - Chapter 10 (section 10.1b)[/li]

[li]Jawad & Farr – “Structural Analysis and Design of Process Equipment” Section 12.2 (see page 424 second edition)[/li]

[li]Troitsky (Lincoln Welding Institute) – “Tubular Steel Structures – Theory and Design” (second edition), Section 5.5[/li]

[li]Harvey - “Theory and Design of Modern Pressure Vessels” – doesn’t seem to address this issue [otherwise, about the best textbook on vessel analysis around][/li]
[/ul]

It seems to come down to needing to contact the ASCE report's authors for an explanation why this was omitted in the new edition. It may be because the current edition doesn't really seem to care how the loads on the anchor bolts are determined, it's more concerned with how the anchor bolts are embedded in the foundation.


ETA: the only texts that give a fully comprehensible discussion of this method, IMO, are Brownwell & Young, and Troitsky. I would never have been able to understand it from just the Jawad & Farr text or Megyesy.

 

[JStephen]

Tom, thanks, I missed it in Bednar, and the discussion there is actually more in-depth than in Megyesy. He does give references for it there (dating back to 1944).

In the derivation, he assumes strain is proportional to distance from the neutral axis. Which is based on plane-sections-remain plain. And as I recall from beam theory, that is true away from supports or discontinuities. So it seems to violate the assumptions used in its derivation. I don't know that that makes it any less accurate than the alternative method (which assumes the same thing). The obvious solution is to put strain gauges on some anchor bolts and see how reality compares to the different derivations. I've not heard of that being done, and would be curious if any of those references had any additional justification for the method.

In some of the current tank standards, they have requirements for prestressing of bolts, generally only applicable to higher-strength bolts. So Gr. 36 bolts aren't prestressed, but others are. I assume they have special equipment just for this, but haven't dealt with it.