Seismic Analysis of Diesel Tanks

[asp1111]

Hello All

I need to analyze a Genset fuel tank for seismic loading per ASCE 7-05 and certify it for a certain g level.
I am applying the full hydrostatic pressure on the tank wall times seismic g accl (with the required load factors) on the tank walls and after analyzing am getting high stresses.

Is there a less conservative way to do the analysis?
Any help is greatly appreciated.

Thank you.

 

[JoshPlumSE]

Check out ACI 350.3: Seismic design of liquid containing concrete structures.

There is also a PCA Publication called Design of Liquid-Containing Concrete Structures for Earthquake Forces. I've found both to be very useful.

http://www.cement.org/bookstore/profile.asp?itemid=EB219

 

[JStephen]

How big and what shape is the tank?

If it is a vertical cyclindrical tank, you could apply API-650 App. E to it.

If it is small, there normally wouldn't be any analysis associated with stresses in the tank. (You don't seismically analyze the gas can in your garage, for example). It seems like ASCE 7 has design loading for rigid structures that could be applied in that case to check for overturning.

 

[asp1111]

Its about 1500 gal. 20 feet x 8 feet x 5 feet approx. rectangular steel tank made up of gauge 7 plates.

ASCE 7-05 talks about storage tanks and API also does the same I believe.

I am looking for any code/document/study which specifically address these type of structure in the event of an earthquake.
The approach of applying full hydrostatic pressure along with the seismic g accl on the tank walls seems a little too conservative and is resuling in high stresses.
Can anyone suggest any alternative approach?

Thank you.

 

[JStephen]

I'm not familiar with the ACI document referenced above. TID 7024, upon which the tank codes are based to some extent, includes an analysis for a rectangular tank, which is then extended to a circular tank. Concrete tanks are sometimes rectangular, so that ACI document may include seismic loading for rectangular tanks.

There are a number of different possibilities to help with the issue:
Current tank codes typically separate loading into impulsive and convective (sloshing) loading. For larger tanks, the convective loading is a reduction from the impulsive. Current methods combine the two via root-mean-square, and only combine a portion of the vertical acceleration with horizontal.

You might look at failure of the tank versus overstress. Flat plates when overstressed may develop some permanent deformation, but that wouldn't necessarily constitute a failure.

You might look at large deflections in the plate, which will reduce the calculated stresses.

If the tank is free to slide and seismic forces are very high, you might reasonably limit seismic loading to the sliding force.

Make sure you're not combining strength-level loads with service-level stresses.