I am having problems in modeling air fin cooler outlet header, and any help will be appreciated.

The problem is that there is a gap of +/- 9000mm between existing (3) and additional new (2) bundles. It is a two phase flow and equal distribution is crucial. This is a revamp project and there is no space to achieve flexibility of the piping.
So I tought that a simple solution would be to make piping very rigid and strong so that the outlet manifold (14")acts as a trust block between new and old set, and moves bundles easily (7500N per nozzle to overcome the friction force).
Trust blocks do exist between 3 bundles of the existing set as well as between 2 bundles of the new set.

The assumption was made that all bundles are loose except the first one on the right of the existing set and all of them are fixed in front where inlet and outlet header box is.
I model the system as follows:

1. restricted the first right bundle of the existing set in X direction and input other calculated X displacements due to thermal expansion at the other five of the existing nozzles
2. left pipes free at the new nozzles in X direction and input friction force Fx = -7500N
4. run and checked X displacement at the first left nozzle of the new bundles
5. went back to input this X displacement, remove the force from all and input other nozzles displacements due to thermal expansion of bundles,
but the restrain forces (Fx) in the output file (outlet manifold) are exceeding 7500.

How to interpret this? That displacements in input file are wrong because manifold is expanding at 68 degC and bundles at 232 degC?
Allowable loads on air fin cooler nozzles are per API 661. Is my approach making sense at all?

Regards
Ranka

The problem you describe is very complex, it is not a simple question. A knowledgeable consultant would require several days to sort out this system.

However, I can offer a few tips and things to watch for.

• Make sure any interaction between bundles (or bundle and header) is addressed properly, i.e. don't assume nozzle displacements unless the nozzle is on equipment. If you assume a displacement (at a location where a branch meets a header), you are ignoring the interaction between the two systems. Depending on the magnitude of this interaction, you could be close to the correct answer, or completely wrong.
• Be careful with rigid body motion problems, if you are counting on large displacements and friction. Remember that CAESAR II, like most <em>first order</em> analysis packages assumes small displacements and small angles. Large displacements and <em>second order effects</em> are not addressed.
• Friction is a non-conservative force - what goes in may not necessarily come out - static and sliding friction are not the same. There is an good writeup on friction in the CAESAR II Technical Reference Manual.
• Along the same line of thought, be careful with the friction configuration, this can change the results. Read the on-line help on these items and play with them to ensure you have a good solution.

I hope this helps.


------------------
Regards,
Richard Ay (COADE, Inc.)




[This message has been edited by rich_ay (edited February 25, 2000).]

[This message has been edited by RSofijanic (edited March 01, 2000).]

Dear Richard,
Thank you for your response. I am still struggling with the same model and still need a bit of advice.
I modelled the existing system as it is, with displacements at the nozzles of air fin cooler. The nozzle loads are exceeding the allowable per API 661. But this system is operating without problems for years.
So I realised that my model was wrong. Most probably by inputting displacements the nozzles were acting as an "displaced" anchor which is too conservative approach.
The header box (where the pipe ties in to the bundle) is not anchored for the bundle frame but the bolt holes are slotted and the pipe moves the header box freely.
Then I modelled the header boxes of the bundles as big heavy pipes resting on +y supports and consequently loads dropped drastically. I also tried to model bundles as a weightless rigid body, and loads are again very low (below API 661 allowable).
Please advise if any of these approaches are correct and safe, and which one is more accurate?
Best regards
Ranka

I think your problem is the correct stiffnes
at the nozzle. If you make a anchor with displacements, then you will become conservative nozzle loads.The best way is you
give realistic stiffnes at the nozzle (modelling a WRC Nozzle or take stiffnes from Finite Element calculation)and modelling the vessel body with rigid element and the real restraints.

The problem is that this is neither cylindrical nor spherical vessel. The header box is rectagular and WRC does not apply. I do not have FEA software.
Please discuss if and why modelling of a heavy pipe in place of the header box is not an option.

And one more thing...

Over all the years of doing this work I have had numerous times when piping, or equipment was overloaded when analyzed, and yet the new work which I had just analyzed closely matched an existing installation which all the knowledgeable people using it said "it works just fine"

So what gives?

In some cases the people saying that everything was fine are flat out ingnorant and wrong. An example of this is an owner who said that his pump seals lasting nine months between failures were fine... whereas actual MTBF for that set of seals should have been 3 years!

In the case of static equipment don't forget that thermal overloads are fatiuge based and will take many cycles to have their effect, and if they do cause a crack its is most likely that the crack will be blamed on bad fabrication as opposed to ****py design.

Also for the final fact to ponder on... every good engineer applies a safety factor to their work. The B31 codes build various factors of safety into our work. The same is true I'm sure of other codes as well. As also is the allowable loadings for rotating equipmment. If the over load is small enough "eating" into this factor of safety has no negative effects but, it may. So a design which is slightly over may work out by blind stupid luck.

------------------
Best Regards,

John C. Luf

Your dilemma illustrates the importance that boundary conditions have on the overall response of the model. I agree with the comment about using specific flexibilities for your nozzle condition. You have limited automatic options here (WRC-297, API-650, and BS-5500), all of which have their own set of rules and limitations.

The <em>finite element</em> option is probably your best alternative. If you don't have access to the necessary software, perhaps assistance from outside could be considered.

Another alternative to check would be the cooler manufacturer. Can they provide the necessary nozzle flexibilities? After all, they gave you the load limits ...



------------------
Regards,
Richard Ay (COADE, Inc.)

To correctly model air cooler,one must understand the equipment. The bundle normally can move if the piping force is high enough. What I did is modelling bundles together with piping system.Input weight of bundle and support friction. If piping is rigid, bundle will move therefore nozzle loading can be reduced. If piping is flexable, bundle not move, and the nozzle loading will be below allowables.


Regards

Alvin Zhu