I have been stumped on this for a few days, and so have my colleagues, so I figured why not give asking you all a shot.

I am attempting to design a “center pile tower” for a temporary grain storage unit. A schematic in elevation view can be seen here. Grain is transported up the conveyor inside the catwalk and dumped into the chute inside the middle of the 80’ tall lattice tower, where it eventually comes out through one of the five openings , forming a pile which is covered by a tarp. The capacity of this unit is about 1.5 million bushels, to give you an idea (a bushel is about 1.22 cubic feet, something like that).

The tower is completely open for the bottom 20 feet; the rest of the tower is cladded with 14 gauge steel for the remaining height (except for where there are spouts).

Calculating the wind load was pretty straightforward using chapters 29 and 30 of the ASCE 7-10. Calculating the grain pressure on the tower, however, is where I am stumped. Or at least I think we are being ridiculously conservative. We wanted to investigate different cases to figure out the worst-case load on the tower (empty, ¼ the way full of grain, halfway full, ¾ of the way full, and completely full).

At first, we decided to take the worst case, being all the pressure on just one side of the tower, which would occur if this pile is full and will be emptied out on one side when a door is opened. My model would not converge unless ridiculous sizes were used.

We then decided to see what would happen if we loaded the tower with those loads shown in the last link, except they would act equal and opposite on all four sides, effectively “crushing” the tower. This time, my model converged except for when the storage pile is completely full—the P-delta instabilities likely are due to the angles buckling. Turning the P-delta shears off yields a very overstressed model.

I have taken the equivalent fluid density of grain as 22pcf per this document , and the point loads as 22 times the depth times the area of the bay, then divide in half for a point load on each leg. For example, at a depth of 60 feet (or height of 20 feet), P=22* 60 * 3.7 *5.5/2=13.4kips. Which as I said, is ridiculously conservative given there have been no catastrophic failures of towers built in the past and this load is seen on an annual basis. And I've been scouring the Internet to see what is out there on the subject, but there isn't much.

Thoughts on this anyone? I know my loads are off by a factor of god-knows-how much, but I have little idea on how to prove it.

Thanks, everyone.