The system here was brought to me in the late stages of design. It consists of a tower and a vertical reboiler connected together by a straight 12” dia pipe. Both vessels have support lugs which are bolted to an elevated structure. The lugs on each vessel are 45” below the nozzles. Since the entire system is at approximately the same temperature my analysis assumes no differential vertical growth.

Temperatures are 300F design and 220F operating. The vessels are both Hastelloy construction. Their centerlines are 80” apart and approx 29” between the nozzle faces. The tower is 42” ID, 1/4” thick with a 1/4” repad at the nozzle. The reboiler is 27” OD, 3/16” thick, no repad.

The piping system between the nozzles is 12” dia, std. wt. carbon steel, PTFE lined and consists of 12"x2" instrument tees (4" face-to-face) at each nozzle and a 21” long spool piece between the tees. The system was modeled in CAESAR II from nozzle face to nozzle face with appropriate horizontal displacements at the terminations.

Obviously, since the vessels are prevented from horizontal movement by virtue of being bolted down, theoretical forces on the nozzles are in the millions of pounds. For process reasons, the client will not allow an expansion joint. As I see it, then, the following solutions are available:

1. Do nothing! Unrestrained, the system would move only about 1/32” at each vessel wall. Given the flexibilities of the shells and other factors the system might safely see out its life. The problem is, who can guarantee it?

2. Leave the bolts off of the reboiler. I don’t like this solution either, because there is no way to know whether the reboiler will actually move given the vague nature of friction.

3. Mount the reboiler on Teflon slide plates. At the present stage of design, this would involve expensive changes: Either the reboiler support structure will need to be lowered or the elevation of both vessels increased with consequent rework of piping.

4. Support the reboiler on hanger rods. Again, expensive structural changes would be required. In addition, other piping attached to the reboiler, not to mention wind and other factors, will create a very unpredictable system.

At present, I am inclined to recommend solution 3. In the meantime, however, any thoughts, suggestions, recommendations or comments from the forum would be highly appreciated.

# 4 is the best # 3 is an acceptable alternate.... but the important thing to do is to let the so-called designers know what a stupid mistake they have made....

The other thing is do not accept blame for the late solution the designers hould have been thinking of this all along!

another possible solution is to model everything in great detail, including nozzle flexibility, support structure flexibility, etc. and see if you really have a problem. Note that you need to make sure you have considered all the transient operating conditions, steam out, etc, in addition to the expected normal operating conditions.

Don't forget to include the stress due to axial loads in your flexibilty analysis. See para. 319.2.3(c). If will not govern over nozzle loads, but it would be the correct way to evaluate the stress.

Ricardo,

The option #1 would be the best approach in my opinion - it is a matter of including the nozzle flexibilities per WRC107/297 to have more accurate understanding and assurance of the stress conditions in the vessels and the piping. Along with the two nozzle flexibilitites, there would be added flexibility from flange joints with PTFE flange faces and any gaskets. A single PTFE envelope gasket with its compressible core would be able to absorb the 1/16" total thermal growth. Two gaskets could be better. If you want to include a very small gap for piping fit-up, then the cold spring effect could offset the theorectical forces of millions of pounds. Since the short PTFE lined spool does not accomodate piping fit-up tolerances between the equipment nozzles, you might want to include a solid PTFE spacer or two for adjusting by thickness or taper.

The probable two phase flow out of the reboiler would 'hammer' an expansion joint, so it is understandable that the client does not allow an expansion joint. I would avoid an expansion joint for almost all corrosive services.

The designers did the right thing to support the two equipment items at a common elevation. There was another separation column and reboiler of zirconium that had the reboiler supported 30 ft up in structure. The column was skirt supported at grade. Vertical differential growth between the reboiler and column would have required an flexible expansion joint - there would not have been sufficient space for large piping loop. It was early enough in design that the column was revised to be lug supported and structural steel was revised to carry the weight of the column and reboiler at the same elevation near the top of the reboiler. The reboiler had a vertical outlet with elbow over to the column for a small amount of piping flexibility.

Chuck's solution is the only real option. The others are assumptions and guesses or at least they will be until you sign the calc. "Well, we had an expert look into and he said it was correct". Admittedly, all your approaches would probably work in practice for 99% of the time. If they didn't, gasoline would be $20/gall. The bottom line would be as Harry once said, are you feeling lucky.

Following Chuck's advice might just work. Its better than the current ifs and buts. It'll give you more confidence because at the moment you don't know. It will also give you a more authoritative argument if it doesn't.

A slight flexing of the steel and net loads from all other connected pipe might work with you.
Then again, they may not.

If the most realistic calc doesn't work and doesn't win the argument, your ONLY fallback option is to cut your losses and leave. You'll be forgotten about in no time. There is always somebody out there who will sign the calc. Don't get into a bun fight. You are definitely NOT going to win an argument if they can't or won't understand the simple facts. You will then be black balled and guess what - the system will get built anyway.

Best Regards

I vote for Chucks option but remain dubious about the where that journey will end.

However even a stiff nozzles slight rotation will drastically reduce moments.... The steel may also provide some reduction

In the past I have seen this approach work sometimes and not work other times...

Mr Weighhell and Mr Becht have my vote.
For my twopeneth,
It is common practice to refine your model and calculations as required until you have a acceptable answer to your analitical problems, and this problem is a good example of when a good accurate model is required.
80" between centerlines will give thermal growth of around 3-4mm. which i suspect the nozzles will struggle with.

with regards to ptfe gasket flexibility, i brought the same issue up on this forum a few years ago, but ptfe flexing is a (Stochastic?) impondrable. and something that cannot be quantified in an argument.

ideally, the two units should be rotated 90 degrees to allow for a small loop between them,

my suggestion is this.


1.Indicate to your 'Higher Authority' the problem and hold any further design if possible.
2.Model as accuratly as possible.
2a.Be realistic in the analysis
3.Formulate solutions in order of cost, time and practicallity.
4.Leave the final decition to the owner and indicate any (possible)risks with the chosen solution.

This is scary when only condiderations have been made about the modeling of the system but little though about pratical side, fabrication of the spool to exacting tolerances and how to install the spool. How are you going to seat the gasket? Have you looked at the stresses in the flange when you try and pull the vessels togeater to seat the gasket? This is all based on precision during fabrication which does not happen. Even if you had the spool welded up in place for the best fit possiable you would still struggle to get the gasket to seat and give you a leak tight seal with out over stressing the flange and bolts. Your only solution is to tell the truth that they have an unworkable design.

William,
I'd be the first argue that you need to consider fabrication tolerances and other such details in stress calcs. In my nearly 30 yrs experience, your not going to be very popular or employed if you do.

One example of many happened a few years ago when I witnessed a 12" nozzle neck snap and move 1/16" away from a cooling water pump located in the bottom of a "dry" sump. The pipe was virtually fitting to fitting. The Emergency stop was also in the sump and could only be accessed by wading knee deep in the leaking water by then touching the 415v motor junction box. (Not by me) I wrote to the head of piping suggesting we take fab tolerances and the other layout factors into account on future jobs. He replied saying he checked the layout, pump type and stress calc. All was correct and hence it was not a design problem. It was also non of my business. I don't work there or on staff anymore.

You can take a donkey to water but you can't make it drink. You'll be kicked if you do.

Reality is always a stern taskmaster... however construction people can work to tight tolerances or no tolerances depending upon their skills and what drives them kind of like the people doing this work.

All one can do is assume that the construction will be up to the task and if it isn't its obvious.

By the same token constructability needs to be accounted for.

Your bolt holes on the equipment alone are going to be about 1/16" oversized to begin with.

That said, I also support Dr. Becht's call for a more rigourous analysis. The structural engineers should be able to give you some guidance on stiffness of the support structure and a quick run with FE/Pipe Nozzle Pro will give you some stiffness values for the two nozzle/shell interfaces.

For a short a run as you've got an the temperatures you're looking at, I wouldn't be surprised if your loads drop dramatically.

You are missing the point I was trying to make about using a program to solve a problem which is not just a modeling problem. Caesar is a great tool but it is only a high tech calculator which will not get you out of poor design layouts. Common sense and the old saying if it doesn't look right it probably isn't.

William,
I agree 150% but this view is not wide spread.

In my experience, Caesar is often used to justify less than adequate layout esp in these days of 3D. By the time info is extracted from the model, it is far too late to change. Persuading designers to give up prelim "hand drawn" info to confirm layout in principle is the subject of another post.

Compare the PM as a lawyer asking a question. He does't want a "yes but ...". "Is it right or is it wrong?". "Well if its not completely and totally wrong then it must be right". We might know there is a grey area between a mile wide. As Confusious said, "the calculated stresses and nozzle loads are within design code requirements based on the following assumptions".

If you are going to be detailed, you must go for Finite Element Analisys. Even if the equipment is big, you are better off modeling with FEA software than with CAESAR II. I would suggest to use FE pipe since is already preprogrammed to deal with vessel and pipe configurations. Try modeling everything in the same model/software instead of transferring data from one software to the other. Since you are looking for a fine detailing even the smallest thing such as the deflection on the equipment might help, and you could probably miss these small effects when transferring data.
Regards,

Here is another reason why FE Pipe could be a good approach.
Recently I had to solve a problem (justify adequacy) of a nearly straight pipe, with only two 30 degree bends, which as expected was creating high anchor loads. I used FE Pipe to calculate flexibilities of bends as well as anchor plate using MeshPRO, a new add on to FE Pipe, hoping to reduce this high load. The net result was a very marginal reduction of anchor load from 8.5 to 8 KN. However, in the process I discovered that the anchor plate could actually take in excess of 10 KN, a lot more than the specified limit of 6 KN, which in my case solved the problem.

I agree with Andrew Weighell that CAESAR II is often applied only to justify adequacy and because of that is not used to its full potential. That was precisely the reason why, back in 1997 we developed PDMS to CAESAR II interface via CADWorx/Pipe. It was to enable quick access to the design model, carry out early assessments before it is too late, perform iterations and analyze “what if” scenarios, in other words to become proactive and to become integral part of the design process. As a result, over the years we managed to eliminate design errors like the one mentioned here and optimize our designs, which in the end saves much more than what people usually see in an overvalued saving in modeling time.

CAESAR II is a beam element program. While it is a very good beam element program which is more than adequate for the majority of needs there are occasions when FEA is a better if not indespensible tool.

When and hows are a tough call but I have seen FEA fail to meet some hopes and sometimes exceed other hopes as well. In the case you cited you hoped FEA would decrease the load but you found out that the beam element solution was a good solution for the anchor load.

"I wonder if you decreased the elbow K for the 30 Deg Elbows?"

Anyhow both tools have a time and place but most important are the skills of the person handling the tools.

The B31 codes are based on the simpler beam element method (i.e., they do have the depth of advice as found in Section VIII Div2). Do a search on the COADE newsletters I wrote an article some time ago on this subject matter.

Final thought beam element analysis is far less time than FEA even if you are using FE Pipe.

Misa,

The critical path is NOT the time to model the data. It is getting data in the first place. To get the interfaces to work as the book intends, you need data in the PDS/PDMS model which means lines need to be complete and with pipe supports located. Most times in order to prove a design is adequate in principal and to avoid the situations above, you only need rough "scaled" dimensions which are quite close enough for 99% of pipe stressing. I am not against 3D, only that stressing should be done before the model is fully detailed. Then again, its the same rate for doing a job the second time as it was the first.

From yesteryear... a true flexibilty analysis by hand methods was weightless (excluding Spielvogel who based everything on the centroid of the system)...

A quick weightless T only run with no supports will quickly show the relative flexibilty or lack thereof of a sytems geometry. (This will not provide compliance with the B31 codes and should be used only as a quick check!)


Remember Markls paper was titled "Flexibility Analysis"

1) I think modelling everything exactly will not solve the prolem as the temperatures are quite high and pipe size is big.
The only feasible solution can be mounting the reboiler on PTFE as mentioned by El Gringo.

2) If you can not change anything due to cost implications and sure that the design will definately work practically, then try to find the same kind of system which is already in existence and working satisfactorily since many years. I came across the same kind of problem and I also thought in the same directions. However client had same kind of 4 trains already working since many years and so don't want changes in that. The question was how to prove the design on paper ?
So I quoted ASME B 31.3 clause 319.4.1 (No formal analysis required) and excluded the system from the analysis. However let me clarify the that particular system was not
so critical and had low operating temperatures.

Hi all,

I am trying to grab this opportunity while I believe all the participatant in the subject are already considered experts (special mention to Chuck). Not very often I heard or encounter the expression "it works" and my first reaction is that it did not or is not failing. Overstressing in computer analysis simply mean to me that any such figure is above the code alowable - the magnitude however could have a serious implification to the system or system components actual failure. For example 99.99% passed and 100.1 failed but what is the difference? - very negligible of course. The standard set it at 100% and over means failed. On the above scenario, and I'm referring to Mr. Rayjaguru's item no. 2, when he said a system will work okey does it mean okey because it did not show any failure yet? And so here is my question: since over 100% failure in the analysis does not realistically reflect a pipe's failure (if it really does when?) can we let the analysis go on reds (code required combination of load cases failed maybe 150% or more) if we feel that the probably similar exisitng system has worked for sometimes? Thanks and God Bless this forum - to me the best mentoring I could ever get. More power to all of you.

Ed Lamigo

Explaining red ink and the phrase "Code compliance failed" to the client is not a path i like to follow. Yor primary purpose as a stress engineer is to ensure the design is as per 'the' code, in a language the client can understand. I can assure you, Project managers can become hysterical at the very mention of non compliance.

which is why i would not readily use ASME B 31.3 clause 319.4.1 (No formal analysis required)
unless i had the original calculations and could prove the two systems were identical.

Guily until proven innocent comes to mind.

Superpiper how EU of you "Guily until proven innocent comes to mind"...

The question of what is or is not a code overstress once came up amongst the folks in TG B. Some people said 99.99% of the code allowed values were okay, some said anything over 90% was overstressed. My guess is an RFI would say anything less or equal to the allowed design stress is acceptable per the code. However B31.3 has a significant amount of caveats in it (as does B31.1 to some different extents). Anyhow basically both codes are predicated on the notion that for the most part thermal displacements impose bending stresses. (Indeed B31.3 ignores torsion unlike B31.1 a fact that I feel should eventually be changed to include torsion).

The codes do not have any interaction equations in them to limit compressive loads. The system as described that we are talking about probably has large compressive loads. These should be taken into account despite the fact that CAESAR II may very well spit out a report that says alls well you are still the responsible person who needs to determine when and how to go beyond the simple beam analysis results....

see http://www.coade.com/newsletters/jun00.pdf

I am also not of opinion to use "No formal analysis required clause".

As I mentioned that system was not critical and with low temp. By "It Will work" I mean that when know thermal growth is hardly 1-2 mm and that can be taken by gaskets, nozzle and shell flexibility.
And sometimes you have no choice particulary when nobody including your PM and client want any modifications. When client shows you the exactly same system working since 10 years how can you ask for costly modifications into that?

Sometimes we need to look beyond books.

fatigue damage may take more than 10 years to appear... so the fact that a system has operated for 10 years without failure is hardly adequate proof, especially if the system has a lot of strain energy built into it.

John, B31.3 does include torsional stress in flexibility analysis, however, it does not intensify it.

Also, compressive stress would be required to be considered per para. 319.2.3(c) when it is significant. However, you will also have to realize you may need to check column buckling.

Use of 319.4.1 is problematic. Remember, though, that you as the designer are responsible for your judgement in the application of that paragraph. If the piping or nozzle fails, it is your responsibility, and you cannot point to that paragraph as blessing the system. It only gives you the latitude to exercise judgement and does not make the judgement for you.

I would consider it bad practice to rely on gasket flexibility. First, the gaskets are already compressed by the bolt up. Second, if the system thermal expansion is actually able to significantly compress them more (which it will not do until the system thermal load exceeds the bolting preload), it would likely result in a leaking flange joint when the system cools down.

Chuck, Yes so sorry I meant to say we have no SIF’s for torsion (haste makes waste)….

As for this thread and its contents it goes to show the state of affairs we all find ourselves sometimes in.

Speaking from first hand experience in a system that does contain raised face flanged bolted joints where the gasket is required to carry the internal bending and compressive loads the system may pass code displacement stress checks and compressive buckling checks and yet be an eternal maintenance nightmare in service. Most raised face flanges and their gaskets are not really designed for carrying large-scale mechanical loads.

An owner I was working for tried to assure me that a system carrying huge compressive loads due to thermal strain worked just fine as it was. When I did all the checks for buckling etc. it seemed to work out just barely. But when I asked about the flanged joints and the service records for the piping system.

It turned out the flanged joints leaked so badly in service that birds flying overhead died in the vapor clouds leaking out of the flanges. It was such a common problem that bird catch screens were erected over the joints to keep their carcasses off the flanges. (this with class 1500 flanges mind you)

Somehow it just didn’t seem to be an appropriate design to me and I added flexibility to the layout… after the request I made for written direction from the owner that allowed the leakage to occur was turned down.