Design method for multiple saddles

[vesselguy]

Hi All,
It's been 10 years or more since I last did a multiple saddles design for horizontal vessels. I will soon embark on one soon and just like to see if anybody know if there are any new (or better) methods or paper on this subject.

The method I used was to first model the vessel as a beam on an elastic foundation. With the help of the civil engineer, I estimated a spring coefficient for the foundation. Calculate the reaction forces for each saddle. Use that force and manually apply the load in Zicks analysis (taking a leap on simplication and assumptions).

What is everybody doing now for manual calculation for this kind of problem? I don't want to hear FEA; not everything needs FEA.

Thanks.

 

[Estassoc]

When I am designing very long pressure vessels with multiple saddle supports (manual calculations), the continuous beam equations are my favorite tools to calculate all the required forces and moments acting on
each of the saddle supports.

 

[MJCronin]

Vesselguy et. al......

The methods currently in use are still very much the same as the ones used 10 years ago.....

Many people have bought software (from places like COADE) that automate the load development phase...

Some of the pressure vessel texts I have read caution against designing horizontal axis vessels with more than two saddles. The caution is based on the uncertainty of foundation settlement ( i.e. the load ends up on the wrong two supports)

Has this been considered in your design ?

MJC

 

[vesselguy]

MJCronin, Estassoc,

I didn't know Compress do multiple saddles.

MJCronin, yes, whenever possible avoid multiple saddles design due to the reason you said in your thread. However, some vessels like bullets (over 200 ft long) will end up using 3 saddles. This sort of design will require close team work with the Structural Engineer (the guy that's doing the foundation and piers) to estimate the worst saddle load combination base on pier settlement. Now a days with all the structural software, it should be easier to get the combinations.

It turns out the problem that I am suppose to do is now dead and I won't need to do multiple saddle design. I use my MathCAD worksheet and found out the comment made by another engineer was wrong and it won't need multiple saddles. This turns out to be one those situation where the projet manager took the "free" advice made by someone and made a multi-millon dollar decisions without substantiating calcuations.
However, I am finishing my zick program (I build my own engineer toolbox from VB) and would like to add a routine to do a 3 saddle design. Do you know of any paper that discuss the method? Brownell & Young only have one paragraph about this kind of design and it just covers the equation to use for the horn stresses over the saddle. Any lead on information on this subject to read further on is appreciated.

 

[LSThill]

Hi vesselguy (Petroleum)

How are going to handle the Stresses at Horn of Saddles if not using FEA ANALYSIS?

The Standard FE/PIPE saddle model con be used to verify the Zick calculation for saddle stresses, to calcluate stress n the saddle and vessel due to axial static seisic load, or to evaluate the effect of pressure or load cycling. The stresses can show indicate that Zack may have underestrimated the bending stresses at the horn of the saddle due to his use of strain gage for maximun stress predicition.


vesselguy (Petroleum) Look at Photo Gallery(Stress at Horn of Saddle):

http://www.thill.biz

Leonard@thill.biz

http://www.thill.biz

 

[LSThill]

HI vesselguy (Petroleum)

do an Advance Serach: Saddle

Leonard@thill.biz

http://www.thill.biz

 

[LSThill]

Subject: vessel bending stress on saddle wear plate

Regarding Attached Stress Plots and Saddle Horn Stresses:

This introduces two problems that we've tried to resolve with the FE/Pipe shells:

1) Singularity at concentrations that are mesh size dependant
2) Concentration vs. Discontinuity stresses

Stresses can often have a characteristic of both concentrations and discontinuities, and can have linear components as the mesh refinement gets better that tend toward infinity. Certain geometries and element types display this behavior more readily than others. Norsok 004 attempts to address this issue and I think that you'll start seeing more
of this discussed as the Code bodies try to embrace FEA on a logical
footing.

Norsok discusses actual element mesh sizes and the tendancy for stresses to go to infinity and recommends element sizes for discontinuity stresses that are on the order of 1/2 the wall thickness. The Code should definitely be reviewed to see the application of this guideline. I mention it here because that particular element size for bricks is often thought to be too large. The problem is further complicated by the actual geometry of
fillet welds -- which is unknown. (Again discussed in Norsok.)

At PRG our approach to date has been limit Pl - which tends to be less sensitive to these problems, and move the bending stress into the secondary catagory. This is particularly justified in the case of saddles and horns when the vessel weight load stresses do not cycle. (As they typically do not.) We think that this is born out when one compares the Zick horn stresses to FEA results. In many cases the Zick bending stresses at the horn are seen to be the FEA membrane stresses.

Dependable Code compliance using general purpose FEA tools, (and special ones too) definitely requires a thorough look at the problem.

That said, for the model you've shown, we would follow the guidance we feel is implicit in the Zick approach, and look at the membrane stress and compare it to 1.5Sm. For the membrane stress, more elements through the thickness often HELP the singularity problem, because singularity stresses
are nodal only, and more elements through the thickness isolates the stress artifact over a progressively smaller and smaller area.

By singularity, we mean that as you refine your mesh the stress will continue to go up. Unfortunately, for the model size you have, doing this will probably be prohibitive for you to see the result, and so you just have to hope that the mesh gradient you stopped at produces the desired "highest stress" unless you take steps to minimize the effect of the singular stress problem

Regards, Tony Paulin

 

[vesselguy]

Leonard,


You have a very nice web site. Your name sound familiar, did you attended the ASME Div 2 course in Orlando in year 2000?

If one is to consider cyclic effect, then yes, FEA needs to be used. I can't afford to get FE-pipe as I am an independent contractor. Only the larger companies I worked for can afford it. Most of the smaller firms I have worked with don't even have Compress or even exchanger thermal design softwares like HTRI or HTFS. Hence, I have to get back to manual calculations like it was back in the 80s. It worked in the past and no reason it won't work now.

In Brownell and Young, it suggests that for multiple supports one should use L equal to 2x the length of load carried by the saddle. If L>= 8R use the formula in Eq. 11.42. for the circ stress at the horn of the saddle. This, as I recall, is what we used when I last did a 3 saddle design over 10 years ago for a 250 ft long bullet. I think we all understands that Zick, WRC 107,etc., are "in the ball park" estimate type manual calculations. I would look at how close the resulting value is to the allowable and judge whether I should pursuit more detailed method such as FEA.


I worked for a large EPC and their US Principle Engineer also recommended this practice, as I recall. This company's practice is to use spring loaded supports at each saddles to compensate for some pier settlement in the future. It's been so long that I can't remember some of the details and methods. I sure would appreciate some tips in this area by the experienced guys.

Take care.

 

[vesselguy]

Hi Mr. Paulin,

Wow we got the famous Mr. Paulin on this site! I meet you in your FE-Pipe course in your Houston office last year Feb. Loved the course. Only if I can afford your software.... I also want to thank you for your great software support in the past.

First off, what is the reference "Norsok"? Is it some paper a guy name Norsok wrote? Any idea where I can read it on line somewhere?

I know this is an overkill but I'm just throwing out a thought. If the circ. bending stress in the saddle horn is an area of uncertainty, won't it be better to just put in an internal stiffener for this case of multiple saddles and make a contact angle at least 140 deg or larger? This way, we won't have to deal with the horn stresses and vessel can be considered as an uniform beam. NO?

I know that one most likely need to use FEA for cases when one has a narrow width saddle, dual saddle stiffener rings adjacent to the saddle, or high temperature (over 600F)service. But for most butane product bullets the temp is low and at the end we all just want to design a vessel that we know would be safe. We might oversize some things for safety but considering the cost of vessel in the light of the big project cost, it is nothing but a drop in the bucket. The total piping and valves cost are way more than normal temp vessels. Those project guys don't give hoot about our techincal issues (or refinements), they just want to know if we finish designing it and have we meet the schedule.

Now is 2002, there's got be some paper somewhere with close form equations that are available for multiple saddles design. I don't have to work on this problem now, but it is a subject of interest to me. I'm sure other people around here is interested in this topic as well and would like to know.

 

[LSThill]

Hi vesselguy (Petroleum)

Yes, vesselguy (Petroleum) did you attended the ASME Div 2 course in November 13, 2000 Orlando, FL.

Also, FE-Pipe course in your Houston office last year January 2003, and CODEWARE COMPRESS Jan. 2003.


Leonard@thill.biz

http://www.thill.biz

 

[MrVessels]

I have a photocopy of a paper written by several professors at the National University of Singapore that is entitled, "A Simple Solution to Cylindrical Pressure Vessels on Multiple Support".

I have no direct experience of applying the methodology described therein but can make a pdf version of the paper available if you wish.

e-mail me at cxslater@bechtel.com.

 

[LSThill]

MrVessels (Mechanical)

Yes, Thank you for the best technical paper on: "A Simple Solution to Cylindrical Pressure Vessels on Multiple Support".

Leonard

 

[PEHasan]

In my opinion, we should avoid providing more than 2 saddles as it will not allow the vessel to expand. When we provide 2 saddles, one is fixed while the other one is sliding.