To all forum members,

It is the old topic : stress analysis of lines connected to Air Coolers.

My personal perception of the problem is: The Air Cooler being a much stiffer structure than the connected piping , will not be governed by the piping movements , rather the piping will be governed by the air cooler movement. An exact modelling , simulating the stiffness of the air cooler is not possible , hence I would prefer to model the cooler like this:Each header box is modelled as a rigid structure with +/- displacements at the two ends ( equal in magnitude ), connected to the header pipe.The displacement is computed using the alpha *Delta T *L ( L= Half length of the header box along its axis).The supporting of the air cooler is by +Y with friction, restraint along the direction of tubes ( for fixed header boxes) and restraint along the axis of the header box ( i.e. along direction of L) , with gap.

There is another school of thought, which I don't agree with, is the use of linking arrangement between the header boxes ( i.e. connecting the air cooler header boxes by structural member), to get them as a connected unit, to take care of differential movement between header box and connected piping.

I would like to get the opinion of experts in the field ( John Luf, John Breen kindly take out some time)on the above linking arrangement and also if somebody can share their experience with me on this arrangement and name of a vendor who has done this type of arrangement.

Thanks and Regards


Anindya

I have met the same problem before, I prefer second modeling which I think is more realistic. Take care of the cooler supports, which normal four legs each cooler. Otherwise, result is totally difference.

Anindya

Some of the engg firms , consider the nozzles as anchors and proceed with the analysis. Obviously these piping configurations are very flexible ,at that high elevations ,requiring extensive supporting structures.

The other school of thought which most of the established engg firms follow , probably takes lead from the API661 notes ---" The application of the momemnts and forces in Table 4 will cause movement that tend to reduce loads..."

The header boxes are understood to be pushed by the forces generated by the connected piping and hence will NOT generate fixity. The Stress engineer is supposed to closely coordinate with commodity engineer from the RFQ stage , to realise this through the use of "Thrust Blocks " or "Low friciton slide pads" at the Header box supports.
The Thrust Block term used above can be misleading to a Stress Novice , but is nothing but a structural member used to connect header boxes.

Bottomline--

The 2nd school of thought yields a compact design solution . The Responsible stress lead engineer has to decide on the modelling approach.

As you said a there is certainly Judgement involved in Air Cooler piping stress analysis..

I am sure the peers in this forum will further enlighten us on the basis of this approach..

Neelam,

Thanks for sharing your views.

I somehow could never agree with the API 661 statement of ---" The application of the momemnts and forces in Table 4 will cause movement that tend to reduce loads...".

Reason: The stiffness of the piping is much less than that of the header box assembly. Also coupled with the fact that generally the piping header and the header box assembly are connected by flexible piping. The piping header may be stiff enough, but the two elements( header box and header are connected by flexible member).So I feel that the API661 statement is not valid for most Air Cooler arrangements.

Things might look allright when we do an analysis considering the linking arrangement, but there also the results will show marked difference depending on what size we use to model the "rigid elements" for the header box.Basically in the linking arrangement we are considering that the piping arrangement will make the header box move. How do we substantiate that?

I have seen in many consultancies the approach of "nozzle movement" calculation in analysis of piping connected to vessels. In this approach basically we are not considering the stiffness coupling between the piping and the vessel , i.e. a combined model approach. I agree with this view that the vessel movement will not be governed by the piping movement and hence there is no need for a combined model.If I try to draw an analogy between this method and that of my approach to air cooler analysis, a counter argument might be: for a vessel we have an anchor situation, where is the anchor for a air cooler header box?My response will be that considering the mass and the frictional restraint to the header box, the near anchor situation in case of the fixed header box, are we not having a situation similar to that of the vessel? Why then have a different approach for the Air Cooler?

Also I have found many analysts model the restraint boundary condition at the Air Coler supports , not only with translational but also wit rotational restraints.Argument, it is not possible for piping to "rotate" the Air Cooler. If that is the case, how can it make it translate?Translational stiffness being much higher than rotational .

Ironically all the different approaches work i.e. I have not ( with my limited experience) came across cases where Air Cooler nozzles failed due to piping loads, which would have otherwise failed due to different modelling approaches to the problem.

I would like to get some more views on this and also request somebody to kindly give me reference of vendors who have designed air coolers with linking arrangement.

Regards

Hallo to all members,

Went through all your opinions about modelling of air cooler in Caesar II and thrust block arrangement. As an experience of the same I am sharing my corresponds with air cooler vendor.

We are doing engineering of a hydrogen plant located in India. During engineering we came across stress analysis of air cooler piping. During initial stages of analysis the nozzle loads were to high because of very stiff piping connected to it. To reduce nozzle loads we have planned to do the piping as flexible as possible. Then we lend up in very big loops and supporting that loop piping becomes very difficult task. Then we tried mechanical linkage arrangement. The cooler consists of 6 bays. We fixed the inner supports of middle bundles and outer bundle are connected to each other with mechanical linkage (Thrust block). The air cooler is modeled with actual weight as given by vendor. Then we communicated the same arrangement to vendor with the displacements at each supports of outer bundle. Accordingly they have provided the thrust block connection. Also they have shared their experience on the same subject. The vendor has already supplied such kind of air coolers to other projects in India (Well known Indian refineries). I have the information related to air coolers supplied with mechanical linkage arrangement. The air coolers in previous projects supplied by that vendor are working well since its commissioning in 2003.

Waiting to get some more views and comments on this subjects.

Thanks to all.
______________
Adinath Miraje

Hallo all,
Sorry to re-open this old thread again. Modeling of air cooler is discussed in his forum many times. I have used both methods for modeling air cooler that is mechanical linkage method for multi bay air cooler (4 or more bays) and first method described by Anindya(with additional rotational restraints ) for single bay or two bay air cooler .
But this is first time that I have a split header even pass air cooler. Looking at detail drawing ,I am not sure whether to include inlet and outlet piping in same model.
Since the air cooler has only two bays I have not used mechanical linkage method for modeling air cooler. My problem is how to simulate split headers.
Since inlet and outlet header boxes are not connected to each other, I have modeled inlet piping and outlet piping in separate model and have used total air cooler weight in both models to get conservative result. I would like to know if my technique is correct to simulate split header air cooler.

Khalidmf,

When you say split header box, to you mean their is a physical game between the inlet box and the outlet box and they are on the same end of the air cooler?

Edward,
Yes, both inlet and outlet header boxes are on same side of air cooler and are physically separate.
As per API 661 ,If the fluid temperature difference between the inlet and the outlet of a multi-pass bundle exceeds 110 Deg.C (200 Deg. F), split headers are used.

Regards,
Khalid

It's been a long time since I've come across a true split header box like that as they rarely get designed for a 200°F+ differential.

As I recall, I would pipe both the inlet and outlet sides up tightly so that each could push on the header box to slide the finfan bays outward. I would model the inlet and outlet separately.

As always with a multibay finfan, you will want to check the movements at those outer bays to see if the installation clearances need to be enlarged by the vendor.

Edward,
Thanks for sharing your views.
As suggested, I have already modeled inlet and outlet separately. Also vendor has provided sufficient Gap for Header box movement.
One more thing I would like to mention is that I have considered Inlet and outlet header as independent headers to check and qualify sum of all nozzle loading on header.

Regards,
Khalid