I’m currently modeling a 6”, 5156mm Diameter, Heater Ring. I have 40 branches living the header, so I have to make 40 elements.
Doing so and running the error checker, the following errors appear:

1) Bend Element 170 to 180 is not LONG ENOUGH to contain the bent curvature.

2) Bend Angle on element 170 to 180 is 4.5 Deg; it must be between 5.00 and 95.00 Deg

3) BEND ANGLE 3.375 specified for the bend element 150 to 160 is within 5.00 Deg of the bend far end point. Either use the bend far end point, node 160, or move the specified node more than 5.00 Deg, away from the end point.

As I already mentioned the ring has large radius, and I have to make small elements (4,5 to 9 Deg)in order to place the branches and supports.
I went to Configuration Setup, to Geometry Directives tab and made the following changes:

Minimum Allowable Bend Angle 2 (Default 5)
Bend Length Attachment Percent 2 (Default 1)
Minimum Angle to Adjacent Bent 2 (Default 5)


These modifications fix the problem. However, according to the Caesar Manual, can lead to numerical problems, or less accurate results.

Do I have to find out another way modeling my system or can I trust this?

This System this HEATER OUTLET 1500 F HYDROGEN Line so please answer only if you REALLY know what exactly are the effects of these modifications to Caesar Solution Method.

Thanks in advance

"HEATER OUTLET 15000F HYDROGEN" I hope you mean 1500F!

You are correct. 1500 F

see here this may be of some help... http://www.coade.com/dwnld/CAESARII/IMPACT.ZIP

360 / 40 = 9 degrees. It looks like the simple fix for your problem is to make the first of the 40 connections parallel to a principal axis, and then you can increment all the others around and stay away from the 4.5 degree bend error. But this may not match your real world arrangement since your piping may be arranged so that all of the feeder lines have at least a 4.5 degree bend.

I'm not sure why the math routines would have any limitation on bend angle. The B31 Code doesn't consider the arc length of the bend in calculating an SIF, so CAESAR II shouldn't either. If that's the case, then the check for bend angle is purely cosmetic, intended I suppose to protect those of us who work a lot with sloped piping from accidentally messing up the sign on the slope.

I'm not sure I would trust CAESAR II to do a capable analysis of the toroidal header, I doubt that the B31 rules apply to such a structure. I'm also not sure that, even if you think the B31 rules should apply to the toroidal header, the assumptions about bend elements made by the B31 Codes are valid for bends with numerous non-negligible point loadings spaced around the periphery.

I strongly recommend that you use a FEA program to take a very detailed look at a portion of the toroid before you go signing off on something like this. This has a huge potential for making a very loud noise some day.

Kyr,

I don't hink you have a problem. I think you are running up against limits in the Caesar II error checker which are there to prevent incorrect bend modelling. The bend angle limitation is to ensure that you meant the bend defined to be there, and limits may be varied. the Bend Length Attachment Percent is a numerical 'trick' employed by Caesar to ensure that a zero length element can't exist, when for example you define an anchor (not common) attached to an elbow. A small element is added at the elbow end.

I discovered this when benchmarking with Caesar, a third party calculation of a long radius bend with anchors both ends. The orthogonal thermal loads were initially not the same, due to that little element. With a bit of tweaking the results were equal.

cheers,
MoverZ

I guess I would say the following...

The Bend SIF and Flexibility Characteristics in the codes that CAESAR II uses is based upon the premise that a single elbow with straight tangent lengths on either end are long enough to allow the bend ovalization due to an inplane closing moment to occur freely. The paper by Markl entitled Piping Flexibilty analysis covers this topic.

Obviously a toroid is not a single elbow so these code based factors will be less applicable to the rings behavior. How far??? I'm not sure but in general if you restrict the elbows ability to ovalize its SIF and characteristic H will decrease.

The other question of branches and point loads are separate issues. All in all if you run up your code stresses to the max I would not be comfortable with CAESAR II alone

Thank you for the replies.

My main issue is that I have to make an accurate model. I have dialed with high temperatures before (ok not so high, up to 550 Celsius) and one thing I’ve learned is that with little changes, or tweaking in the Caesar model, you can get completely different results. Pipe material is near yield point and it is like a gum. Now with 850 C, I have to be SURE that I have a correct model.

Speaking of FEA Analysis, for me is a world where I’m never sure that I’ve made the correct assumptions (correct element selection, solution method etc) when I have to dial with such high temperatures.

Well, when I’ll complete modeling the system, I’ll upload the ._A file.

Best Regards,

Kyriakos

I recommend you consider using the Hot Mod E for one model to see exactly how your restraints will act at temperature....

You can use the model for code stress check only if you ratio the stresses by the ratio of the Mod Eh vs Mod Ec see B31.3 319.2.2 note 4

I do not know what is your material. At 1500F most have serious creep issues. Check what is the case with your material. For me codes do not address creep design in full. What ever is the material I would not rely only on CAESAR II, code recommendations or even with FEA. Make research and find out what has worked in the past.

Piping Material: INCOLOY 80 HT SB-407 UNS NO8811.

It is not a new system. It is a replacement of the existing line. We will keep old routing, I have to recalculate the hangers. That's why I'm trying to make an accurate model.

I do not want to make just a copy of previous hangers because:

1) We don’t have all the data about the old hangers

2) I’ve learned not to completely trust old designs and "do it as it was" or "it's working so do not change it" thinks because it is hard to find out about malfunctions or problematic situations of old systems (possible failure of one of the hangers, in my case) unless you talk with operators, old personnel of the refinery etc.

I think there is a lot of good advice given above. I am not going to comment on whether you can rely on Caesar II/B31.3 completely for this analysis because I am unsure of all the issues, but if you want to rid yourself of those errors I can help you with that so at least you can run the analysis and draw your own conclusions.

The errors on the length not long enough to contain the curvature of the bend is exactly that. Remember that we code between tangent intersection points so the second element has to be twice as long as the first (when using equal angle consecutive bends). Your first bend element length must be equal to L1 in the attachment. The following consecutive bends must have lengths equal to L2 ( = 2*L1 for equal angle bends) in the attachment. You last bend element will again have a length equal to L1. The other thing to remember is that your direction will change (most will be skewed elements) with each successive element (look at the direction of L2 in the attachment to see what I mean).

To remove the errors related to the bend angle, select Environment->Caesar II Configuration File->Geometry and reduce the minimum bend angle and the minimum angle to adjacent bend.

Here is the link to this article from our newsletter (April, 1995) because the quality of the attached pdf is not good.

http://www.coade.com/company_newsletter.asp

I recommend you read the notes for this material in ASME II D allowable stress table.

No design code is perfect. They cannot cater for all possibilities. It is easy to find examples where design was fully to the code but failed very quickly.

Kyr,

It may not be the 100% correct approach the the problem solving, but it will give a lot more accurate result withous straggling with the modeling; I would use 40 or more straight lines with small radius at the connections of straigh lines. This will simulate almost the same behaviour/weight of the entire ring with 40 branches. I believe it will size the hangers correctly if the only concern is sizing hangers.

Or sometimes running the same calculation with the structural bar element packages may give you an idea about the external load or support load distribution of the piping, the rest is the spreadsheet calculation to the code rules.

I do not want to repeat the advices on the material issue, it was covered already.

Regards,

Ibrahim Demir

Kyr,

As for the getting nodes on the bends, this may work.
Bury the Ring header and then play with the soil stiffness value till you get the restraints spacing as you need. And then delete the bi-linear restraints and use those nodes for the branches.

Might work or not in your case. Worth a try though.