Finite Element calculation for floating roof 11

[Hamidreza1973]

Dear All,

For your kind information, we have some internal floating roof storage tanks in our project. The client insists that we have to use the finite element calculation analysis for this issue -Pontoon and deck- to be modeled by SAP2000. Is it a normal request?

Best Regards

 

[LittleInch]

No

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[JStephen]

It seems to me it would be a rather challenging application, due to large deflections, buoyancy/ varying liquid level, design controlled by buckling, large physical size with small details included, rotation, etc.

 

[HTURKAK]

the client is free to ask FEM analysis.. Is it for API tank ? ..If so, Pls look API 650 App. H for requirements which is normative ( H.4.2 Internal Floating Roof Design ).. In oder to see the deflection and developing stresses , FEM is the best way to go..

I looked to the web and found a worked example ..

 

[Hamidreza1973]

Super perfect..................
Is this software your home development? and do you have a sample of metric (SI) calculation file sample?

 

[Geoff13]

35+ years, and I've never heard of FEM for a floating roof.  I would tell the client NO.  In any case, I'm not convinced a floating roof can be correctly modelled in FEM.

 

[LittleInch]

Ask him WHY he wants a FEM model for something that literally just floats on top of a large expanse of liquid.

There are no real stresses or forces to look at here and internal floating roofs inside tanks have been around for decades without any particular Issue I'm aware of. They tend to be thinner and lighter then EFR's because they don't have anyone walking on them or any rain or snow, but literally just sit there and go up and down slowly.

Just because you CAN analyse something to death doesn't mean you SHOULD or NEED to.

what's it going to prove?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[HTURKAK]

Err. HAMID REZA,

If you provide detailed info. (size, material, type of IFR, support locations , pontoon sizes if applicable,..local loads ..) if the tank is designed as per DIN EN 14015 Annex c...

If you nedd ,I may model your IFR with SAP 2000 FEM and provide SDB file just for random kindness and sharing knowledge..

 

[Hamidreza1973]

Dear HTURKAK and Geoff13,

This is the sample that the Owner provides to us for your kind information.

 

[Geoff13]

This is the sample that the Owner provides to us for your kind information.

I have no doubt that it's possible to make an FEM model that looks like a floating roof (FR).  What I believe is very, very difficult is making a model that acts like a real floating roof.

I will note that I am not trained in how to make proper FEM models (we have a dedicated group for that), but I have worked with the group enough to understand the difficulties.

It is impossible to review the package you attached in any proper way without the FR drawings, as well as the FEM model and software.  Nonetheless a few things jump out at me.
[ul]
[li]Clause 5.1.3 : Why are the buoyancy pressures for the intact roof different for the deck and pontoons?  In my mind the buoyancy pressures are an output of the FEM analysis, not an input to the model.  See also Figure 15 comment below.[/li]
[li]Figure 4 : I have never seen an FR with so many structural members.  Perhaps this FR works differently than ones I am used to?[/li]
[li]Fig 6, 7 & 8 : I've never seen an FEM model with such large plates, and elongated elements. This generally results in poor modelling.[/li]
[li]Figure 11 : The number of boundary conditions should be the minimum to keep the model from translating.  I would have used one or two X-direction springs and also one or two Y-direction springs.  This many fixed X and Y constraints will hide the real roof action, particularly during the punctured deck and pontoon condition.  If the roof weight and buoyancy loads are properly balanced the model should not need to be any Z boundary conditions, other than a light spring since the load balance can never be exactly zero.[/li]
[li]Figure 15 : The buoyancy load looks to be perfectly uniform on both the deck and pontoon.  Depending on the pontoon rotation in operation there will be some variation.[/li]
[li]Figure 16 : With the centre deck and two pontoons punctured there will be notable global FR rotation, as well as varying circumferential rotations, resulting in differing buoyancy pressures both around the circumference and across the pontoon radial width.  As noted above buoyancy pressure should be an output, not an input.[/li]
[li]Fig 35 & 36 : There is almost no change in the figures at the two punctured pontoons.  In fact from these figures it's hard to guess which are the punctured pontoons.  No doubt this is due to the problems noted above.[/li]
[/ul]

This report is about what I would expect.  The customer is happy because they have an FEM report on file, but the contents of the report are utterly meaningless in terms of accurately modelling the real floating roof.

 

[simplemath2]

Geoff13 - a lot of your question can be answered by boundary condition used in floating case:

"When the roof is floating on the process fluid, the deck stays level with respect to the XY plane. Based on that, in order to bound the model in the Z direction, displacement of the deck. nodes are fixed in Z direction. On the other hand, since the outer rim of the pontoons is constrained by the roof seal to the shell, displacement of nodes on the outer rim are fixed in the X and Y directions"

 

[HTURKAK]

Dear Hamidreza,

I do not look every day to the web and regarding for your last post and attachement ( internal floating roof calc.) ,i am not sure if you are asking my comment..

I just througly looked to the model and try to understand the geometry of the IFR.

SAP 2000 is general purpose structural analysis program and capable to do FEM analysis for more sofisticated structures than a IFR. Moreover,SAP 200O plus version can do wave analysis.

My points,

- Apparently, the designer has done seperate Buoyancy calculation and loaded the IFR for for intact case ( i could not see for the punctured case)

- The reference frame , if X an Y are horizontal, Z would be vertical direction and the nods SHALL BE free in Z direction,

- The ring type pontoons would be composed of ( bottom and top deck, bulkheads , inner and outer rim plates, stiffeners ) ..that is OK.. However, the use of stiffeners for single deck at middle will KILL the membrane behavior and meaningless ..

- I do not know your position for this project and not sure for the reason , Client's demand for FEM analysis. But , the client has the right to ask FEM model analysis to see the developing stresses . My suggestion would be , ask the services of an experienced engineer on SAP 2000..

My Opinion...

 

[Geoff13]

"When the roof is floating on the process fluid, the deck stays level with respect to the XY plane. Based on that, in order to bound the model in the Z direction, displacement of the deck. nodes are fixed in Z direction. On the other hand, since the outer rim of the pontoons is constrained by the roof seal to the shell, displacement of nodes on the outer rim are fixed in the X and Y directions"

I realize this is a quote from the FEM report and not your own comment.  I hope you won't mind if I explain my disagreement with these statements.

Operating Condition
The pontoon area is heavier than the deck area, and thus would like to sink further into the product.  However they are joined together and thus the outer region of the deck is dragged down by the pontoon, and the inner edge of the pontoon is lifted by the deck.  It may be possible to specify Z-restraints in the centre of the deck, but not all the way out to the pontoon inner edge. A simple check of whether the Z-restraints go too far out is that they should have close to zero load in them.

This lifting of the inner edge of the pontoon will cause a general rotation of the pontoon, which will cause the pressure to differ from the constant pressure specified in the model.  This increased pressure at the outer edge will try to reverse some of the rotation caused by the deck, and the final rotation will be where these two effects balance out.  This is why I suggested the buoyancy pressures are outputs (from this balance) rather than fixed inputs.

Punctured Condition
With two adjacent pontoons punctured the pontoon region will experience a global tip, which will further sink the pontoon area, and drag down some additional portion of the centre deck.  The Z-restraints that worked for operating may need to be released during the punctured condition so they remain essentially at zero load.

When the centre deck is punctured its self-weight will now pull inwards on the pontoons where it is attached.  This should generate a circumferential compression in the pontoons, along with a torsion if the line of action is below the CG.  However if XY-restraints are placed around the outer rim this centre deck pull will go to the restraints rather than the pontoons.  Again the check would be to look that the restraint loads are close to zero.

The seal between the FR and shell only provides just enough push-pull to keep itself filling the gap.  It is incapable of preventing the pontoon area from moving radially in the same way that FEM restraints permits ZERO movement.  Thus fixed XY-restraints can't be used.  If XY-restraints are placed on the outer rim they would have to be light springs with a value matched to the actual FR seal pressure.

 

[JStephen]

In typical plate-design equations like in Roark, they assume small deflections relative to the thickness, and that assumption is necessary in order to keep from developing tensile forces in an otherwise flat plate. In a typical floating roof, deflections will be "large" relative to the plate thickness. In addition, that large deflection will change the geometry, change how deep the roof floats, change the stresses in all the plates, etc.
In a typical floating roof, the plate dimensions are such that they don't very well fit into standard beam-bending/ beam-buckling criteria. It is possible to model buckling of something like that in FEM, but it's not necessarily something that is automatically generated, either.
In both those cases, I assume a FE model COULD be generated to capture those effects, as well as the tilting/punctured condition mentioned above, but I would check very closely to see if that's the case. And more to the point, if major effects like that are neglected, there's not much point in a FEM solution, which brings you back to the posts above.

 

[Milad -H-CH]

Dear Hamid Reza
please find the sample sap calculation for normal float condition for internal floating roof in attached file.
best regards

 

[Milad -H-CH]

Dear Hamid Reza
please find the sample sap calculation for stay on leg for internal floating roof in attached file.
best regards

 

[Milad -H-CH]

Dear Hamid Reza
please find the sample sap calculation for Twice dead load condition for internal floating roof in attached file.
best regards

 

[Hamidreza1973]

Dear All,

Really thanks from all of you to discuss this issue. I really gain from all of you and so many thanks.

Another point: I attached the PDF file of the buoyancy calculation. As you can see, Owner commented on this calculation. What is your opinion about his statement?

Is it better to increase the outer rim size to reduce the plate wasting or reduce the rim size so we can extract two pontoons from one plate 6 * 1.5? (Please consider also the cad file attached to the PDF).

 

[HTURKAK]

The client (apparently supplying the steel plates ) in this case fully right. The sizes of pontoons more than overkill..Some rule of thumbs;

- If the wt of IFR is 125 tons, the total volume of pontoons would be 250 cu-m would be OK,

- The width of pontoons would be 2.0-2.5 m..

- Adjust the rim hts to minimize waste ..

If i were, would prefer pontoon width 2.0 m...

 

[Hamidreza1973]

Dear HTURKAK,

thanks for your reply. The owner already supplied the pontoon plate 6(m) x 1.5(m). With this consideration, is the attached PDF a better solution or you have any other opinion?