I am presently carrying out an analysis on an air cooler piping system for the first time. In this system the inlet header is resting on the cooler inlet nozzle. When I modelled the piping system with only the nozzle displacements on their mating flanges, I observed that the Forces, Fx, Fy and Fz are pretty high even in the sustained (SUS) case. I was wondering if I could model this as a "Y" restraint with the displacements at their respective connecting nodes instead, since the cooler nozzles would act as a rest for the inlet header in the sustained case. Please I would like to know your views and any advice on this issue.

Thanks.

I usually model the air cooler header box as a set of rigid elements going to a common anchor point. Your piping header and air cooler header box should be at essentially the same temperature, so you should not get loads due to differential nozzle movement.

In cases like this, I much prefer to construct a model of the equipment instead of inputting calculated displacements. When you have two or more connections like that to the same equipment, even though you've calculated the movements "correctly" I find that the nozzle movements won't match the piping movements enough to avoid Caesar reporting significant loads. The loads do not exist.

pipepro,

I tried yesterday to reply to your post but when I hit "Post Reply" it disappeared.

I know the problems you are having, I've been there. Because your system is so tightly coupled, you're not going to be able to displace the inlet nozzles or header boxes without getting the high forces and moments that you have experienced. Your loads in the sustained case are caused by the slight elongation of the header due to internal pressure. A simple fix is to set header temperature to ambient and pressure to zero but be sure to consider your design temps for the vertical branches and vertical header box displacements.

Another approach that I use more often is to simulate the air-cooler header boxes with a model using rigid elements. Just model one header box and duplicate it for as many as you need. This approach is also good if your oulet nozzles come off the bottom of the same box as your inlet nozzles. Attach your inlet branches to the header box model with full anchors and Cnodes. Restrain each of the header boxes for X, Y, RX, RY & RZ. Allow freedom in Z only. I usually fix the center bay(s) of the air-coolers with a Z restraint, limit stops or friction. As the header expands the header boxes will move appart. To minimize moments, pipe the inlet header and branches as tight as practicle. If the outlet nozzles share the same header box, include the outlet piping system in the model also and provide flexibilty as required in the outlet branches.

Let me add the following tips:

1. If your air-coolers contain a lot of bays, check to make sure you have the available clearance for the header boxes to displace, especially the outer most. Most air-coolers have 1/2" clearance each direction. You can get extra clearance by pre-positioning the air-cooler in the frame so all the clearance is on one side. If you need more than an inch you'll have to ask the vendor to provide the required clearance. I usally ask the vendor to pre-position the coolers as required and lock in place with shipping bars.

2. Ask the air-cooler vendor to provide low friction slide plates to minimize the forces required to slide the header boxes.

3. Request that your allowable nozzle loads are in addition to any loading required to displace the air-cooler header.

Good luck,

Kevin,

I would guess you are describing the principles detailed in a 1981 design guide that many stress engineers carry. ie. Close coupled inlet headers with low friction header boxes.

Please accept my apologies in advance if I have misunderstood. It can be subject about which stress engineers have very strong views.

This procedure has puzzled me for many years. It has led to countless hours of debate in the best traditions pipe stress.

The following comments relate to muli-bundle units. I am refering in particular to the units at the extreme sides subject to the greatest amount of header expansion.

Many refinery air fin exchanger bundles are up to 30-35 ft long and weigh up to 5-8 tons. In order for the 2 (say) inlet nozzles located at one end to be pushed and pulled by the inlet pipe parallel to the header box, the whole unit has to move perfectly squarely. ie. The other end up to 35 ft away has to move in perfect harmony. The friction loads of the bundles, even with low friction bearings create a high moment load at such a distance from the nozzles.

Also, depending on the construction, the tube supports are not low friction even if the header box supports are. ie. the bundle side walls are fixed to the support structure. The only "moveable" parts if they were able to overcome friction are the headers and tubes.

Further, as hard as I look, I have never managed to find an air fin tube bundle that gives any indication that it is moving as predicted by the frictionless floating bundle theory implied in many pipe stress calcs by the use of deflections only along the header box.

I have recently visited one refinery where extra bundles had been added alongside existing bundles. One set of bundles had been piped with a "close coupled" fitting to fitting assuming frictionless bundle movement. The adjacent indentical bundles were piped up with a loop between the header and each nozzle assuming limited or no header box movement. I've been on sites where different units have been piped up based on different philosophies but never before seen it on a single multi-bundle unit.

I am now stopping to put on my hard hat and take cover.

Andrew

It's a good thing there is alot of Factor of Safety in equipment and piping design.

The "Truth" is proably somewhere between the two extremes i.e. fixed movment by the nozzle versus a frictionless resistance.

A note of observation on mu, first even with TFE slides mu cannot be depended upon with any certainty. I have seen equipment "hang up" on TFE slides (my guess is due to dust or dirt). Once kicked loose the slides seemed to be fine but for the first displacement things may not slide as neatly as predicted.

What to do??? Well ask yourself what happens if mu works against you. Will substantial damage occur before things slide? Or will the equipment simply experience an overloaded low damage cycle? Will you or some other knowledgeable observer be around for the first cycle?

If mu can break the damned thing I assure you if you have my type of luck, mu will work against you.

IMHO mu should always be viewed as a hinderance rather than a help.

Just one mans opinion....

Andrew,

In my reply to pipepro, I attempted to:

1. Answer his question on why he was getting such high loads.

2. Give him a brief summary on how I have modeled similar air-cooler problems.

3. Provide a heads-up on some items that may not be considered by a first time air-cooler analyst and can cause problems if overlooked.

You have indeed misunderstod if you thought implied "frictionless" air-cooler bundles that move squarely in perfect harmony. We all know only portion of one end is going to slide. I addition to the friction forces (usually un-known magnetude) required to displace a portion of the air-cooler there will be the force required to spring the air-cooler tubes the displaced distance and that thier will be some rotation of the header box.

It's not an ideal world and I agree with John that the truth is probably somewhere between the two extremes. The method I described, while not perfect, has resultated in many satisfactory installations. The analyst needs to match the sophistication of the computor model to what's really required to achieve a reasonable anaylsis of the system.

There are other methods as well, such as employing thrust block/rods but that's a whole new discussion.

For me the biggest key to a succesfull installation is early communication with the vendor about how you want the air-cooler to behave. i.e movements required, clearance needed, low forces to achieve movements, control of movements (limit stops) and your pipe must be able to accomplish this movement. These requirements are usually different for each bundle and I always demand that the requirements are shown on the drawings so there is no mis-understandings.

I welcome the response and critique of my post but would also like you to post how you address this kind of analysis.

Best regards,

Kevin,

Again, sorry to have misunderstood. In reply to your last post, my personal approach is absolutely 120% with all John's comments.

"If mu can break the damned thing I assure you if you have my type of luck, mu will work against you.
IMHO mu should always be viewed as a hinderance rather than a help".

Real life is somewhere between but the "optimistic part" must be treated as just that. 1000 tons/hr of nat gas at 1500 psi or whatever you are handling could make a mess.

The calc must reflect what is likely to happen on site. If I had a free hand, the only way I can see to make air fin calcs totally predictable on site is to fix one end and one side of each bundle. ie.
a) One end of the inlet header box would be fixed with an x, z restraint.
b) The other end of the inlet header box would be guided thus fixing the axis of the header box.
c) The "other end" of the bundle would be guided on the same side as the x,z restraint.
d) The corner diagonally opposite the x, z restraint would be a simple rest.
Pairs of bundles would be supported with the above arrangement mirrored to put the "fixed sides" of each pair together.

I would the build enough flexibility into the pipework to take up the expansions fixed by the above arrangement.

The guides and stops must be substantial and rigid. A bit of 10 mm plate lug with 30 mm hole and loose M20 bolts on the tube side of the header box as supplied on some units is as much use as hens teeth for both guiding and stopping headers boxes. What about the nozzle loads on the tube axis which put the header box in torsion ? Stops and guides must be more rigid than the tube bundle to take any load. What happens to a fully floating bundle during seismic ?

The manufacturer might say his nozzles can take one or more multiples of API. The question you should be asking is can this load be carried through to the side walls and supporting structure rather than just adding to the loads on the tubes.

A Ceasar calc that includes the individual tubes can produce some extremely interesting results for some air fin designs with a couple of multiples of API loads if you have a free evening.

Finally, I would then change the calc back to close coupled header pipe + freely sliding header box when told to do so.

At the end of the day, as drafters used to say in the old days - "you are paid the same rate for rubbing out".