Storm Shelter in Host Building 1

[WesternJeb]

Our firm is designing an ICC 500 storm shelter (250 mph zone) in the basement of a two story building. The shelter takes up roughly 60% of the building footprint (~100'x100'). Due to the large footprint and lack of floor space from the architect, our plan currently is to anchor in purely gravity columns to the top of the storm shelter for the floor and roof above the top of our shelter. The columns would ideally be above our shelter walls, and use some very shallow embed. anchors that would "rip out" in a major wind event.

We are in a high seismic region as well, so we have to detail the base of the gravity columns to withstand the loads imposed from seismic drift and not fail, but be able to rip out during a tornado. Does this seem possible?

Does anyone see an issue with putting gravity columns on the roof of a storm shelter and designing them to rip out during a major wind event? We are also designing for host building collapse using an energy dissipation theory from chapter 2 of Blodgett's Design of Welded Structures, because I feel like someone is going to mention it.

 

[bones206]

We are in a high seismic region as well, so we have to detail the base of the gravity columns to withstand the loads imposed from seismic drift and not fail, but be able to rip out during a tornado. Does this seem possible?

Rather than ripping out, if the anchors are designed to fail in a ductile anchor yielding mode, then it can possibly meet the seismic design criteria and then in a tornado it would continue to yield/stretch until overturning collapse occurs or the anchors ultimately fracture. The key would be providing enough overstrength in all the other failure modes to guarantee an anchor yield mode. This would be my preferred approach.

 

[bones206]

You may want to detail the anchors to provide generous stretch length, to maximize the ductile elongation and minimize likelihood of sudden fracture. Also choose an anchor material with high ductility.

 

[WesternJeb]

I think that is a great idea, Bones206. My only hiccup is I likely won't have enough edge distance to develop the concrete fully to where the anchor controls.

 

[bones206]

You will probably have to detail a robust anchor reinforcement scheme, including lapping long anchor embedment with vertical rebar, and enclosing the anchors in a cage of ties.

 

[WesternJeb]

I believe that would make sense if it was a base of a moment frame column in the lateral system. You believe this approach would be better for a gravity only column that is subject only to incidental loading from seismic drift, though? I appreciate the advice, just wanting to make sure we are on the same page.

 

[bones206]

I believe so. In my mind you want predictable and ductile failure behavior in an extreme event, so that the concrete shelter structure is protected from damage by creating a fuse in the anchors. I use similar detailing philosophy with blast design, and tornadoes are not all that different than a blast event.

 

[WesternJeb]

That makes perfect sense to me, unfortunately. I appreciate you taking the time!

 

[RobertHale]

Be careful about providing a fully ductile connection. Section 304.9 of ICC 500-20 requires you to design for the maximum force that could be transmitted. You could try and let the connections of the beams to the column be the wind "fuse," but you may be in a catch-22 where you are going to have to transmit a crap load of wind force into the shelter because of your seismic requirements

 

[WesternJeb]

That is a great point as well. I considered making the base plate the weak link in the puzzle and only transmitting as much force as it could handle, but that makes me feel a little too uneasy. That would just contribute to the yielding conversation above and distribute the forces somewhere else, beyond difficult to accurately predict.

 

[bones206]

I think by definition if you have anchor reinforcement designed to develop the anchor yield strength with overstrength, then you have designed the shelter side of the equation for the maximum force that can be delivered by the connection.

 

[bones206]

I've loosely thrown around the term overstrength, so I should probably clarify. Since the actual anchor material likely has a higher yield strength than the nominal rating, the strength should be bumped up to account for that so the strength isn't underestimated. In ACI 318-19, Section 17.10.5.3(a) covers this scenario for tension by factoring the yield strength up by a 1.2 factor (along with other requirements to ensure a ductile steel failure mode).

For shear, I don't think ACI really covers this scenario for non-seismic capacity design. I would also consider applying an appropriate material overstrength factor like 1.2 for shear to make sure the anchors fail before the anchor reinforcement.

 

[WesternJeb]

Thank you for clarifying that. My original intent was for the column and building above to completely separate in a tornado via concrete breakout / rupture so I am still trying to go through everything that would entail creating a fuse in the anchor instead.

 

[bones206]

Yea, I think the concrete breakout approach would technically violate the “no damage” clause for the shelter building. I can see the logic in accepting that level of damage however, if you could somehow guarantee the overall structural integrity and function of the shelter is not affected.

The other tricky part of the concrete breakout approach is that the equations underestimate the contribution of rebar in the concrete member (by design, for conservatism). So the actual failure mode might be unpredictable and not what the breakout numbers say. Might not be the clean breakaway failure that we imagine, unless you detailed some kind of sacrificial, bond-broken and unreinforced concrete pocket around the gravity column anchors…

 

[SWComposites]

Yes, trying to design a structure to actually break at a defined load (a “fuse”) is very difficult. One must know (and somehow control) the actual material properties.

 

[WesternJeb]

I Agree. My other concern is the concrete curing to a high strength than I plan; it is difficult to predict and quite a risk. Plus all of the post-installed anchors have so many safety factors built in that they are not a reliable source of predictable failure. The predictable anchor rod yielding might be the best approach.

 

[DanKile]

What about setting the column on a roller, similar to bridge supports? That way you only transfer downward loads. If you need some lateral support, you could design some U-shaped brackets to carry the load from column to pin to base.

 

[WesternJeb]

Hi Dan, I believe there are OSHA requirements that columns must have at least 4 anchors. They also have to be designed for a worker hanging off the side (300# at some eccentricity) and it would tip over if on rollers. I am not sure if there are special exemptions, but those are some code provisions off the top of my head that I know would prohibit such.

 

[canwesteng]

Put nuts on the anchor bolts for erection, then take them off after erection.

 

[canwesteng]

Or use a shear key for lateral loads, and very small anchors for erection, so that the failure load of the anchors is smaller and easy to deal with.