Splitting Header Box for an Air Cooler

[yl0525]

Could anyone there elaborate for me clause 7.1.6.1.2 in API 661 (6th Ed), see below,
- If the fluid temperature difference between the inlet & outlet of the multi-pass bundle exceeds 110 degC, U-tube construction, split headers or other methods of straint relief shall be employed.

For example if I have an box-type multi-pass air cooler, says 4-pass or 6-pass, whereby the difference between inlet & outlet operating temperatures is more than 110 degC.
1. How should I analyse whether the header is required to be splitted, and where to split?
2. Should we analyse the temperature across each pass partition, instead of looking at the overall in/out temperature of the whole unit?

Tq.

 

[SNORGY]

You could probably model this with several FEA programs or CAESAR II with a few creative assumptions. You could probably get results close to reality by drawing a free body diagram and doing the calculations by hand. The things to look at are the different thrusts imparted to the headers arising from the different temperatures in each tube row. It will probably be more of a tube-to-tubesheet joint strength issue than a material yield issue, although tube buckling might also come into play. To quantify at what state you need the split header box design, I would be inclined to prepare a tubesheet mock-up and conduct a pull test to measure the force required to pull the tube out of the tubesheet. This force is likely to be higher than that required to "push" the tube out of the joint the other way due to the flaring of the tube end when it is rolled. I would vary the amount of wall reduction in each of the tubes in the mock-up and plot the pull force against wall reduction to establish what wall reduction gives optimum strength. Then you have a baseline force to compare to the calculations. In any case, you will probably want at least two grooves in the tubesheet for the tubes. If I recall correctly, optimum tube-to-tubesheet wall reduction is somewhere around 10% for most CS designs and probably about the same for austenitics. When I did a similar test for a martensitic (410 S) mock up years ago, the optimum was closer to 5%. Regardless, if your calculated loads come close to the joint failure load measured in the lab, you might want the split header box design.

Regarding the post above, I apologize to the list. I have no idea what happened, but suddenly my half-finished post was published before this one. If I can delete it, I will; otherwise I will flag it as inappropriate.

Regards,

SNORGY.

 

[yl0525]

Hi Snorgy, thanks for your useful info.

Now I am more concerned about the code compliance i.e. API 661 para 7.1.6.1.2.
In fact I am confused whether the temperature difference of 110 degC is refering to the difference between the inlet & outlet (final) temp of the air cooler (for instance, temp difference between inlet nozzle and outlet nozzle)? Or it shall be evaluated at every crosspass instead?

It could be easier to discuss based on an example. A cooler consists of 6 tube passes with below conditions.
1. It is to cool down a fluid from 210 degC to 55 degC.
2. Temp difference at every crosspasses, e.g at front header, the inlet temp of 1st tubepass is 210 degC while outlet temp of 2nd pass is 130 degC, which means the temp difference for this cross pass is less than 110 degC.

Assuming FEA analysis is passed for this header. Per API661, is the header required to be splitted due to the first condition that overall temp diff exceeds 110 degC?

 

[thermmech]

API 661 tries to help us out by forcing us to split the boxes that have the temperature difference between top-inlet and bottom outlet compartments greater than 110­°C (200°F). The temperature difference is for process fluids. The actual metal difference for top and bottom tube row will be less than 110C. But, you can tell that in the majority of the services, if your inlet-outlet temperature differential exceeds 110C, you are likely to NEED split (or multiple-split) header boxes.

The actual analysis should be done for mean metal temperatures of the top and bottom (and all intermediate ones) tube rows. You have to examine normal operation, turndown, low ambient temperature, etc. You can calculate the amount of thermal expansion for each tube, and see how that translates into the load as (elongation of the tubes)/(tube length) x Ey eqauals stress.

That stress can be converted into the axial tube load by multiplying it by tube cross-secional area. Now you can compare this load with tube allowable load calculated by ASME. You don't need to do the moc-up unless you are cutting it really fine and you want to get more out of your design.