Roof Structure Finite Element Analsis of Cone AST

[jmas]

Hi Everybody,
I've recently engaged to evaluate finite element work of another engineer who designed a roof structure for a 44m dia. and 14.6m height cone roof AST. Merging nodes which was in boundary of roof plate and underlying rafter causes erroneous results from him. So then, i analyzed (linear static analysis) the roof without roof shell elements but applying its correponding weight and efferct of wind, live, ... loads in different load cases. i have reached to 53 mm defection for the rafters, considering a central column. the stresses are within acceptable range. But there are following questions:

1- The roof is modeled in a "no-deformation" condition, before applying loads, but during erection it will be deformed before welding other segments together. How far i am from actual situation?
2- What is allowable deflection for the rafters here? i mean "L/?".
3- I tried for buckling analysis but all of my effort failed (with errors in software), considering different modes of buckling analysis in COSMOS/M. My friend also faced with similar situation in ANSYS environment. Is it necessary to analyze it for buckling?
thanks in advance.

 

[JStephen]

API doesn't specify an allowable deflection. So far as I know, allowable deflections in buildings usually relate to ponding on flat roofs (not applicable to a cone roof) or to serviceability issues with cracking of plaster, etc.

Lateral buckling of rafters is usually checked by applying AISC or similar steel codes. Within certain limitations, API allows the rafter to be considered braced by friction from the roof plate when live/snow loads are acting, but requires a separate check with dead load only, not considering the rafter to be braced by the roof plate. If you are using only a center column, you may be exceeding the depth of rafter that API allows to be considered braced by roof plate also.

I believe the most recent addendum added a requirement for checking unbalanced snow loads, which would have a considerable effect on the column design with a single center column in a large tank.

 

[jmas]

Thank you JStephen for your reply.
If we have stresses in the range, but large deformation in the roof,could this lead to excessive flexibilty of roof? I mean consider a future case when a repair worker enjoyed jumping up and down (like trampoline), until whole structe collapses.
I can't find anywhere in AISC codes which discuss lateral buckling of this type of network structures (web of rafters and girders). Do you have any reference?
"...If you are using only a center column, you may be exceeding the depth of rafter that API allows to be considered braced by roof plate also...", which part of API discussed this?
FYI, the tank will be erected in a warm climate, so there will be no future snow.
regards,

 

[IFRs]

If I remember correctly, the maximum rafter spacing equation was based on a ponding criteria. I don't remember what it was though. AS JStephen stated, when designing roof structures, the cone roof plate laterally braces the rafters except when the rafter is greater than 15" tall but not for the dead load only case (API 650 5.10.4.3). So, you have to check the unbraced length for dead loads using the full length of the rafter. AISC clearly has considered unbraced lengths for their beam tables. Large deformations under man loading would suggest a very weak structure - something must be wrong in the design.

 

[Motorspirit]

It sounds to me as though your FEA model is too complicated and ended up giving you an erroneous result. Keep It Simple. you dont need to model the roof plate elements if it is a supported cone roof. Just model the load imposed by the roof plates. Keep you model as simple beam elements.

Although not strictly specified or required I find sticking to a span/250-300 range a practical rule of thumb.

 

[jmas]

Thamks Motorspirit for your reply. Please tell me what do you mean by "...a span/250-300 range"?
FYI, the problem arised when already designed roof structure started to build at site where it is found that single beam of IPE180 (means H=180mm) have a ~20 cm deflection with length of ~21m. So, all the calculation became doubtfull. In this regard, it is asked to check it again. After cheking, it was found that stresses are beyond and by some modification (adding central column and doubling girders in some areas)the stresses came in the range. the question is :
1- is it necessary to perform buckling analysis for the roof structure?
2- if so, what would be the safety factor for it?
Regards,

 

[Motorspirit]

Sorry, I meant for use as a service deflection criteria you should start by trying to limit the deflection to a span/250-300 range.

In your case you had a 200mm deflection on a 21m span ie you have 21,000/200=105 < 250 -300 so you have potential problem. I find this range generally seems to correspond with an accepable strength design.

With regard to buckling analysis. If you mean "do you have to check members for buckling capacity?" - Answer- yes if they are compression members.

If you mean "do you have to perform a buckling analysis in FEA?" answer- No - Keep it simple use tried and tested compression member design from your local codes as has already been stated in this post.

 

[jmas]

To Motorspirit:
How you can find compression force in every single members of a cone roof structure composing of central ring, radial rafters and circumferential girders (by hand calc.), Having different loading, dead load, ...?