Circ. Stress at Horn of Saddle 5

[rkdpv]

How can the Circ. Stress at Horn of Saddle be reduced?
Actual= 227.41 MPa Allowable= 135.31 MPa

Otherthings are fine.

Shell length= 15 m
Shell Dia= 3 m

Thanks n Regards

 

[TomBarsh]

The stress formula from the Zick analysis looks something like this:

S =-Q/(4*t*(b+1.56*Sqr(Ro*t)))-3*K*Q/(2*t2)

Assuming that you are fixed for most variables such as:

shell thickness, t
shell radius, R
saddle load, Q

That means you can fiddle with the saddle width, b, and the saddle and wear plate angle in order to affect Zick factor K. So if you increase the saddle width and/or increase the saddle and wear plate angles you should be able to reduce the calculated stress.

If this step doesn't reduce the stress sufficiently (and it seems like you have a long way to go to get this stress down) then you'll have to try adding a stiffener ring, or move the saddles closer to the heads so they act as stiffeners, or increase the thickness of the shell.

 

[Mandeep1]

good response Tom.



best regards,
Mandeep Singh

 

[rkdpv]

Thanks Tom

Very informative.

 

[racookpe1978]

I've "never" recalled seeing three-saddle tanks.

What's the logic behind that: Unequal and uncontrolled division of weight between the three saddles?

 

[jte]

Three saddle vessels exist, worked on one recently. Seen some (more than two) saddle tanks as well. In general we're taught to avoid them for fear of differential settlement, as I think you implied above. A vessel on three saddles distributes its load predictably. Until one end settles down a bit at which point the load starts to transfer to the middle saddle. If that end settles enough, the entire weight of the vessel is now on the center saddle, usually three times its design weight. Similar stuation may occur with a slightly bowed shell.

So... if you run across a three saddle design, ensure that the foundation is monolithic, and even then, be suspicious that the loading may not be what you expect. A two saddle weight distribution is far more reliably known.

jt

 

[starrproe]

jte,

Three saddle vessels exist, worked on one recently.

It is turning out that I will also need to work on a vessel with 3 saddles in some time.
Can you please share as to what approach did you use for anaylsing vessel on 3 saddles.Somebody from my office suggested using STAADPRO.Since I have no experience dealing with such vessels,can you please suggest how to tackle this case?

Starrproe

 

[SnTMan]

We have been replacing a fair amount of old HX's lately and many of them have 3 supports. To find the reactions I have a beam program I sometimes use (ETBX). The company in the past has used what I believe is called the three moment method via an in-house program, the basis and use of which is not clear to me. More recently we have gone to FEA, which seems to have become our solution of choice for even fairly simple stuff.

Results of the three methods do not seem to me to agree very well.

Regards,

Mike

 

[jte]

Starrproe-

I used a rather large model, one which I would only have dreamed of a decade ago. Used SolidWorks to build a quarter model including contact conditions for the wear plate to shell and baseplate to pedestal (the vessel had an unusual baseplate / pedestal setup). You might be able to do something like that with shell elements, but I used high order solids. Without getting into details, I'll just point out that I was a bit perplexed by the result until I ran each loading individually and realized that their combinations resulted in a situation which I did not anticipate but was reasonable.

I guess I'd say there's more going on here than I anticipated, even as a somewhat experienced FEA / vessel design guy.


Mike-

I'd suspect that the FEA - if properly done - is most likely the better result, but clearly some "back of the envelope" hand calc's should back up the order of magnitude of the results. Wondering how your beam program and FEA models incorporate the weight / stiffness of the tube bundle and tubesheet. Seems that the relative deflection of the (cantilevered off the tubesheet) tube bundle vs the shell may be playing a role. [Have you considered writing up a paper regarding your analysis methods to be published / presented at the ASME Pressure Vessels and Piping Conference? Rumor has it it'll be US central east coast next summer...]

jt

 

[SnTMan]

jte, I would agree, FEA should be the most accurate. Unfortunately, it is all done by one engineer at another location, and I can accept or reject it, but I cannot say much of anything about it, as I am not at all skilled in that art, plus for output I get a pretty colored picture.

Regarding my beam program, it does not account for tube bundle stiffness, it just treats the exchanger as a tubular beam with distributed loading, no doubt a source of error. The in-house program I'm sure does the same.

This discussion well illustrates why more than two supports should be avoided if possible. Two supports, reactions found in seconds with a calculator, more than two, uh-oh...

As for papers, what I've seen is not nearly formal enough to write up

Regards,

Mike