Big square nozzle on a conical roof 1

[orama]

Does any one know where to find the best approach to calculate the stress and deflection on a big square nozzle placed on a bin conical roof. I don't have FEM software.

Thanks

 

[butelja]

I think FEA is your only hope. I've pursued this case as well as large cylindrical nozzles not on the tank centerline, and have been unable to come up with anything analytical. WRC107/WRC297 seemed the obvious approach, but tank roofs are outside of the range of D/t that they cover. If you find something, please announce it loud and clear, as it would be very useful.

 

[Tankman650]

There is a very conservative way to do this for stress. Determine the cone radius at the location of the nozzle. Use WRC107 sphere calculations based on a sphere with radius 2 times your cone radius. (You can derive this based on rotating the nozzle axis on a cylinder so that the local nozzle axis is 45 degrees from the cylinder axis).

Remember stresses from WRC107 are 45 degress from your roof principal axis.

Deflections are not handled in WRC107. You can find some information in Roarks formulas for stress and strain that cover radial loads on a sphere.

 

[butelja]

Tankman650,

How do you rectify that the D/t ratio for most tank roofs is beyond the end of the curves in WRC107?

 

[Tankman650]

butelja,

The D/t limitation is for CYLINDERS not SPHERES. By using an analysis of a sphere you can get around this limitation.

 

[butelja]

Tankman,

OK, but a small portion of a CONICAL roof is CYLINDRICAL, NOT SPHERICAL. Is there published work that states your opinion that substiting SPHERICAL for CYLINDRICAL is valid?

 

[Tankman650]

butelja,

Actually a cone is a CONE not a cylinder or sphere!

The method I described is an unpublished procedure by a major tank manufacturer.

I agree that a cone with a small included angle resembles a cylinder. The roof of a tank generally has a shallow slope and is closer to a sphere. I have and am sure you have also, modeled a dome or sphere in FEA as a series of cone segments.

The problem is to determine stresses in a cone using WRC107 when the paper only addresses cylinder and spheres.

The same problem exists with buckling equations. Most tank codes have buckling equations based on formula for a cylinder or sphere. WRC69 developed a method of equating a cone to a cylinder for these buckling equations. Most fabricators use this for buckling and WRC107 analysis. It does not make the method right, just makes it a way to get an answer(conservative I hope) when none exists.

By the way, I have never seen a published method of using a cone as a cylinder for WRC107 analysis. I wonder how far off the results are if compared to FEA????

 

[matclarke]

Try using nozzle pro from

http://www.chempute.com/nozzle.htm