Steel storage racks

[jay156]

I'm analyzing some warehouse storage racks for a client to give them a load capacity. The problem is, my results keep showing that the capacity is far far FAR below what the rack company says they should be and far below what the client is putting on the shelves already. The rack manufacturer says they should support like 50,000 lbs per shelf, hell I can't barely show them to support 500 lbs per shelf without the columns being overstressed.

This has happened to me every time I try to load rate storage shelves. Are the manufacturers' engineers using some kind of obscure provisions of the code to make their capacities higher, or am I just overlooking something or making a stupid mistake? (I've got the unbraced column lengths set to the distance between the shelves. I even set the K value to 0.5, which I'm not sure is really right anyway.)

 

[DaveAtkins]

Are you using cold formed steel framing design provisions? I assume these racks are constructed of cold formed members.

DaveAtkins

 

[jay156]

Yes, I'm analyzing it in Ram elements using the AISI-16 ASD code.

 

[KootK]

If it's out two orders of magnitude, it's got to be something silly.

Any chance the manufacture's stuff is just the sum of all the shelf "beam" elements in a racking bay and doesn't even include the posts?

For sport, what kind of capacity do you get if you neglect buckling and assume that the posts achieve their squash load?

 

[jay156]

The beams on the upper shelves are okay. On the bottom shelf they're overstressed, but not to the absurdly high amount the columns are. I attached the design results for one of the columns. All the others are about the same.

I assumed a live load of one ton supersacks on 3.5 ft x 3.5 ft pallets, which comes to about 163 psf, with the associated seismic also. That's what the client is storing there now, and they're not buckling, so I'm obviously doing something wrong here. I wish I knew what it was.

 

[JLNJ]

The design of storage racks definitely march to the beat of their own drummer. That's why they have special provisions in ASCE 7.

On first glance, it seems like your gauge might be too thin. Normally racks are made of rather stout gauge material. 950 pound axial leg capacity seems silly low.

 

[jay156]

Well I didn't have calipers, but I did put a tape measure on the edge of the material and the thickness of the post looked like 1/16".

If I increase the thickness to 1/8", the stress ratio comes down to 4.85.

 

[dvd]

You might rummage around to see if you can find ANSI MH16.1-2023, Design, Testing, and Utilization of Industrial Storage Racks

 

[Aesur]

I put this in RSG CFS to run a comparison, if I read the output you had correctly, this thing is 15' tall, braced at 5' o.c. vertically. Assuming you have torsional and both axis restraint every 5 feet I got around 3 kips capacity. Torsional restraint seemed to have the most impact here. Note I had to change the material in CFS, but have the same strengths.

 

[dold]

OP, it looks like you have some big time moment in the column. Are your beams fixed to the columns or pinned? No doubt the manufacturers would be considering a near-zero amount of bending in the columns.

 

[jay156]

Okay, I tried setting the L torsion every 5', but I didn't get much of a difference. Perhaps it's because the beams are transferring moment to the columns as I modelled them as moment connections. I'm not sure they perfectly transfer the moment, but if I release the moments the whole thing is unstable.

This is basically the rack I'm trying to analyze, with just minor differences in the section of the posts and beams, and 3 ft taller.

https://www.lyonworkspace.com/produ...r-with-wire-decking-96-in-w-42-in-d-144-in-h/

Says the capacity of the uprights is 20,600 lbs. I'm getting nowhere near that and I just don't know why.

 

[jay156]

Yes Dold, I was just thinking that too. But if the moments are released, then the whole thing can flop over down its length.

 

[jay156]

Okay, releasing all the moments in the beams and adding a roller in Y up at the top (which doesn't actually exist, but is needed to run the analysis at all) brings my stress ratios down to around 7.83. Still no good, but better than with the beams fixed to the columns.

I don't think it's actually accurate to model them this way though. I'm sure moment really is being transferred into the columns. It must be.

 

[JStephen]

Quick thoughts from glancing through:
I see a 14 ga in there somewhere, which is a bit thicker than 1/16".
Do the channels have a lip at the ends of the flanges (I'm thinking like a Unistrut)?
I thought they showed 48" spacing on the shelves.
They show some capacities per pair of columns and per pair of beams, not per column or per beam.
They spell out that those capacities assume uniform load on the beams, and will be different otherwise.
They mention base plates, so possibly some fixity there.
If you load alternate shelves, you get a moment in the column, but only half the load. If you load every shelf, you get full load but eliminate the moment.

 

[dold]

But if the moments are released, then the whole thing can flop over down its length.

I think that's the tricky thing with a system like this. When you go to assemble these things you can tell that there's a little bit of play in the connection. I.e., the size of the hole in the columns are slightly larger than the pins that go in the holes. So you end up with a connection that is "loose" enough to not transfer the fixed end moments of the beams into the columns in a static condition, but when the whole thing goes to tip over in an earthquake or whatever then the connection kindof binds up as assembly starts to rack and provides "stability", if you will.

 

[Aesur]

The last one of these I bought from Home Depot had a max capacity per shelf but also had a note buried in their documentation that if you maxed out one shelf the others cannot have any loading as that was also the max capacity of the system. Maybe they have something like this for the one you are designing as well?