Deflection of slender thinwall header under pressure only 5

[fifo62]

Dear all



I'm discussing with a supplier about this fact: in presence of a very slender and light thickness pipe header 8" O.D., thk 8.18 mm, 3.8 m length,with four 6" nozzles, designed at 42 bar (g)/170 °C, mat. SB 425 UNS 08825 (See the link to understand geometry) I'm asking for the investigation of the displacement and stress for the overall system with a comprehensive FEA. I'm not confident that a calculation by formula as per ASME VIII Div. 1 and a very basic FEA investigation for nozzles loads by using a small portion of the header in correspondence of one nozzle, can be considered a sufficient investigation. My concern is that, at the equilibrium, the presence of the holes+nipples&nozzles, even with pressure only, may deflect the header in the horizontal plane (X direction whether p.ive or n.ive) not having it a symmetric behavior. I expect this might affects the overall stress of the header and has to be considered in conjunction with any other loads. In case, also nozzles interconnection points displacement vs. piping rigidity might be part of the study. The supplier states that no deflection at all are to be expected in consequence of the non symmetric geometry and material distribution, behavior will be the same as a straight pipe.
My convincement with such a thin header (already bought) is that we have to expect some millimeters of deflection.
A coarse mesh only was provided for the overall header showing that no deflection is acting, but I don't consider realistic that mesh which has less than 20% elements of a mesh properly done (one million elements against more than 5 estimated necessary). The supplier doesn't have a powerful workstation for FEA and a further investigation will cause delay in placing an order to a subcontractor and they want to avoid.
The questions: Is there possible any deflection? There is any theoretical justification to do not expect deformation? Do anyone is reading has an idea If I'm wrong in insisting for the check? Any previous experience? Any literature?

Thanks to all

 

[fifo62]

I feel necesary to add one explanation more to my thread.

When I say: "the presence of the holes+nipples&nozzles, even with pressure only, may deflect the header in the horizontal plane (X direction whether p.ive or n.ive) not having it a symmetric behavior"; I'm thinking to a sort of bending of the header itself.

Thanks again

 

[prex]

No deflection to be expected.
To be convinced of that try to answer this question: what deflection would you expect if, without holes, half circumference of the header was thicker than the other half? Of course none, the centerline would remain straight.
To have a deflection into an arc you need a differential elongation in the two halves, and only a thermal gradient or a transverse load can cause this.
And even if there was a deflection this could be of a really minor relevance, as the corresponding stresses would be deformation controlled quantities or secondary stresses, and as such not required to be determined or accounted for in Div.1.
Concerning the FEM model, what you state is surprising. Back in the age of the mainframes we had models limited to a few thousand elements, but we were able to make good designs, and the vessels designed at that time are still functioning correctly! What I mean is that the correctness of a FEM model is never just a matter of how many elements it contains, it is instead the type of problem, the insight of the analyst and the depth of the subsequent synthesis by the designer what makes good a model.

prex

http://www.xcalcs.com

: Online engineering calculations

http://www.megamag.it

: Magnetic brakes and launchers for fun rides

http://www.levitans.com

: Air bearing pads

 

[trottiey]

I agree with prex on all points, but I'm kind of puzzled as to why you expect bending. I'd like to make sure we've put your mind to ease. Does it help to consider that pressure acting opposite the nozzle is balanced by tension in the nozzle?

Never gauge the accuracy of a FEM by the number of elements. I've seen a clever hundred-element FEM that predated mainframes and gave excellent results - with paper and pencil and a sliderule! At the other extreme, I've seen million-element garbage that took weeks to run and produced very silly results.

 

[DSB123]

I would not consider a 8.18mm thick 8" OD pipe to be thin wall !!! Are you sure it not 8"nb (i.e. the OD being 8.625"). In any case it's not thin walled the D/t ratio is only 26.77!!!

 

[C2it]

You can expect approx 2mm of pressure elongation, axially.

Welding such large connections will cause some deformation due to cooling, but this will probably just bend the header a small amount.

 

[fifo62]

First of all my thanks for your kind replies and sorry for being disappeared for few days but, your replies gave me a lot to think about and were pushing to go deep (I try to do) before to return on the thread.

The example of a "double thicknes" pipe as proposed by Prex is fine because well reppresenting a general deviation form a condition of assial symetry. (which, in the end, is the effect of holes+niples on my header)

What doesn't convince me is the conclusion so I would propose here attached few notes on a pdf file to explain my idea.

I started approaching by the point of view of the theory of elasticity and, If I'm not wrong, the conclusion is that in case of a double thickness body we have a deflection and the importance of said deflection can be theoretically predicted. In case of a true header because of the complexity of the geometry and the material distribution, only a detailed investigation by FEA can give the figure.(particular cases may also give "no deflection")

What is clear is that on a length of about 3.5 m it is quite easy to arrive to deflection that are important (a small numerical example is given in the attachment). The deflection should not really affect the overall stress level for the double thickness body (I agree with Prex). Some complications in stress distribution seem to me are arising by the real geometry of such a header if / when deflecting. This should happen because of the constraints given by the attached process piping (quite rigid in my case, which doesn't allows any "differential displacement of one nozzle in respect of the other as seems to happen on the header). The fin tubes also may give some influence to the general stress conditions because of the imposed rotation and the header/nipples connection and having and axial guide at a distance of about 1.5 m from the header limit.

Seems to me that the request for an overall model is always justified and in general a displacement has to be expected. Do you agree on this?

About the model and associated mesh I need to complete the picture. Load application and also end effects has been checked and were found correct. To stress the model I normally ask to contractors that for any SCL (stress classification line) at least three element are to be involved. With this assumption and considering the volume of the equipment a n° of about 5 million elements is necessary. Do you thing there is anything wrong with this? A coarse mesh as used should anyway be fine in investigating the deformation and a mesh refinement should not change to much the figure; but, on this there are some argues in between the parties (giving detail may cause confusion). What looks important to me is to get a conclusion in understanding if the displacement is possible and has to be expected. Something of interest, being me not trained on ANSYS, should be to know if there is any set up of the program that for sure has to be applied to investigate for possible important deflections. This to increase confidence in reading results.

Thanks to all for the attention, I hope to have some feed back from you all, any comment is welcome.

 

[saplanti]

I would be more interested in the efficiency calculation at the location close to 6 in nozzle and small tube openings. It seems that large opening and small tube openings are in line and therefore efficiency of the header wall might require you to increase the wall thickness substentially.

In case the small tubes (I am just guessing that tubes are between two headers and top header is supporting the entire system) and bottom header supported by the top (this) header, there will be bending moment with the largest value at the centre of top header where the 6 in opening is. From my experience, if there is not sufficient material around the opening there might be buckling ultimately failure under the beam bending.

Hope this helps.

Ibrahim Demir

 

[fifo62]

Thanks Ibrahim
Geometry in between the nozzles and nipples is complicate and looks to be a possible weak point (this is the reason for FEA and to evaluate external loads effects).
With a calculation by formula Longitudinal and Diagonal efficiency for the ligaments was considered as per Fig. UG-53.5 considering the two rows of holes. 6" nozzles and nipples are self reinforcing. Now, I'm sure to understand your point: you mean to consider the Efficiency of the nozzle hole with the niple hole under Fig. UG-53.6? Can you give me few more input on this?

Note - Header is supporting only: pressure / nozzle loads / thermal if any / its own weight / action coming from tubes if any (should be zero in ideal conditions)
Thanks again

 

[saplanti]

Yes, you understood correctly. I was talking about the circumferantial efficiency between 6 in nozzle and the tube opening.
You mentioned that they are all self reinforced, however I do not see that on the drawing you provided. However I would still check the circumferantial efficiency. It may be in the dangerous zone.

It seems that you have an horizontal tube arrangement and this does not introduce large load to consider in the beam bending of the header. In case the nozzle is reinforced against pressure, this will give sufficient material against the bending due the weight action and bending sourced by the probable eccentric trust load at nozzle area due to internal pressure.

Kind regards,

Ibrahim Demir