tube collectors between joist seats 3

[smvk3]

I am designing an occupancy category IV structure that is in a seismic design category of D. The structure is steel bar joists on CMU bearing-shear walls. Where the joists bear on top of the CMU walls, to transfer the diaphragm load to the top of the wall, I am debating whether to use tube collectors between the joists seats per the attached or utilize the joist seats themselves as a rollover load. The question I have is would I need to amplify the load on either the joist seat or the tube collector by the overstrength factor to satisfy the intent of section 12.10.2.1 of ASCE 7-05? Without the overstrength factor applied, the rollover load at each joist seat would be between 2,000 and 4,000 lbs.

 

[jayrod12]

That's already a pretty high rollover load for the joist seats. May want to look at a collector tube for that reason alone. I believe much above 2000 lbs and they start to need to reinforce the seats for rollover.

 

[JLNJ]

I would use the tubes rather than rely on joist seat rollover for load of that magnitude. Without the tubes I don’t see how you could justify the last few deck welds delivering the diaphragm shear to the seat.

I don’t have ASCE 7 in front of me, but if these tubes are between every joist (or nearly every joist) then I think they act as extensions of the shear wall and not collectors per se. You don’t have a singular point of load collection (and possible failure) nor are you “dragging” the force from one location to another. I would not use the Omega multiplier.

 

[KootK]

I like JNLJ's rationale for this and would apply over strength to neither the joist seat rollover option nor the lug option. Neither setup does enough collecting to warrant collector treatment in my opinion.

 

[smvk3]

12.10.2.1 Collector Elements Requiring Load Combinations with Overstrength Factor for Seismic Design Categories C through F. In structures assigned to Seismic Design Category C, D, E, or F, collector elements (see Fig. 12.10-l), splices, and their connections to resisting elements shall resist the load combinations with overstrength of Section 12.4.3.2.

So the consensus is that as long as the tube collectors are placed at every joist space, no overstrength amplification is warranted? With the rollover seat option, there is an edge angle running continuous over the joist seats so I'm not relying on the few deck welds near the seat to transfer the load. The diaphragm load will be transferred into the edge angle then into the top of joist seat directly with welds.

 

[KootK]

I'd be willing to make the same argument even with lugs at every other joist space.

 

[Deker]

So the consensus is that as long as the tube collectors are placed at every joist space, no overstrength amplification is warranted?

Unfortunately I don't believe there is a consensus among engineers on exactly where the diaphragm ends and the collector begins. I've seen both interpretations. A common argument among those who choose not to amplify is that the overstrength factor is meant to apply to discrete connections rather than multiple distributed connections, and that a loss of one distributed connector (in this case, one tube collector) would not compromise the integrity of the load path as a whole.

My concern with that argument is that if each connector is more or less uniformly loaded, failure of one connector essentially does result in failure of all connectors. I tend to base my decision on the ductility available in the connection under cyclic loading. If it's ductile, no overstrength; if it's not, overstength. This agrees with what I interpret to be the intent of the overstrength factor. See this article (and screenshot below) for some additional discussion: Link

 

[KootK]

A common argument among those who choose not to amplify is that the overstrength factor is meant to apply to discrete connections rather than multiple distributed connections, and that a loss of one distributed connector (in this case, one tube collector) would not compromise the integrity of the load path as a whole.

My concern with that argument is that if each connector is more or less uniformly loaded, failure of one connector essentially does result in failure of all connectors.

You didn't say so explicitly but I suspect that your logic there is based upon the classic, unzipping of brittle things. I feel that argument falls apart somewhat when enough of those brittle things are involved and are loaded by a common distribution element. A brittle anchorage failure mode, for example, will be based on the unfavorable tail end of a strength distribution curve. As such, when there is a large collection of such connections, most of them can be expected to have significantly higher capacities than the one that fails first at its design strength.

For example, if you have twenty connections in line and lose one, the demand in the remaining connections goes up about five percent and, in all likelihood, everything remains just fine. I suspect that the capacity at which any potential unzipping would be arrested in such a system is probably well in excess of an amplified demand. The trouble with this line of reasoning is quantifying it. I wait patiently for some reliability theorist to tell me if the magic number is five, ten, or twenty.

 

[Deker]

I see your point, but I think you rightly pointed out that the issue is quantifying it. Your approach could also be reframed as designing for the amplified demand but using a strength reduction factor of 1.0 (or something similar) in recognition of the expected strength (as oppposed to the minimum).

A repetitive connection will of course have more strength by virtue of the number of connections, but I contend that the total demand should still be based on some consideration of the the type of failure expected.

 

[KootK]

Your approach could also be reframed as designing for the amplified demand but using a strength reduction factor of 1.0 (or something similar) in recognition of the expected strength (as opposed to the minimum).

That's right, so long as it's in recognition of the expected aggregate system strength.

Dialing it back to the decision to use joist rollover or not, for many fastening schemes I would consider joist rollover to be ductile. More ductile than most things really.

 

[Deker]

Yes, the aggregate system strength, although I believe that there used to be precedent in the code for using phi=1.0 with amplified seismic load combinations even when checking individual elements. Unfortunately I can't recall where I've seen it.

Philosophically, I don't have as much of an issue being liberal with the strength reduction factor (which I have done myself on occasion) as I do with neglecting to consider the overstrength factor altogether. The effect you described above wouldn't be enough to make up for a 300% shortfall in demand--possibly more depending on the SFRS and how it was proportioned by the designer--unless the demand capacity ratio was kept low to begin with, in which case one could argue that the system already was designed considering amplified loads.

I wish I could offer an informed opinion on available ductility of joist rollover, but where I practice it's routine to use collectors / blocking between the joists.

 

[KootK]

I wish I could offer an informed opinion on available ductility of joist rollover, but where I practice it's routine to use collectors / blocking between the joists.

I'm sure that your opinion would be as informed as anybody's. Even for those of use who utilize joist rollover it basically comes down to one's judgment regarding the situation shown below: metal plates forming plastic hinges with no real propensity for local buckling.

 

[sticksandtriangles]

In my high seismic region (SDC D by a good margin), adding the HSS collector tubes is common and we do not rely on joist rollover in any situation.
I would not "omega"fy the HSS tube collector design, but I would record omegafied loading to the joist girder top chord and then any connection of the joist girder top chord to the lateral force resisting element.

S&T -

http://www.re-tug.com

 

[Deker]

Great picture KootK, thank you. When relying on rollover, do EORs typically specify whether the joist seat will be stiffened or unstiffened? I imagine that would impact the decision on whether it can be considered ductile.

I would not "omega"fy the HSS tube collector design, but I would record omegafied loading to the joist girder top chord and then any connection of the joist girder top chord to the lateral force resisting element.

Not that I agree or disagree, but what is your rationale behind not amplifying the load? Is it the the distributed strength argument? Granted, it's probably a moot point for this particular condition given that the amplified diaphragm demand is trivial compared to the shear transfer capacity of a tube collector with nominal welding.

For those that subscribe to the unamplified demand / distributed strength argument, do you also disagree with the article I posted above that shear studs in a composite deck should designed for overstrength when delivering diaphragm shear to collectors?

 

[sticksandtriangles]

Not that I agree or disagree, but what is your rationale behind not amplifying the load?

Unfortunately I do not have great rationale for why, I deem the joist girder to be the start of my "collector element" per 12.10.2.1 and design accordingly. Anything above that, being the HSS tubes and deck attachments gets the load to my collector element.

For those that subscribe to the unamplified demand / distributed strength argument, do you also disagree with the article I posted above that shear studs in a composite deck should designed for overstrength when delivering diaphragm shear to collectors?

I have not designed shear studs in composite deck for overstrength either... not sure if that is correct. I also do not design chords and their connections for overstrength (last bullet of your table Deker), I only design collectors and their connections for overstrength.

S&T -

http://www.re-tug.com

 

[KootK]

When relying on rollover, do EORs typically specify whether the joist seat will be stiffened or unstiffened? I imagine that would impact the decision on whether it can be considered ductile.

I don't specify it but, at the same time, I have every expectation that it will either work unstiffened or we'll be doing something else (lugs etc). The cost differential between stiffened and unstiffened joist seats can be shocking in my experience. This is third hand information at this point but a very knowledgeable person here suggested to me once that the cost is less about the addition of the stiffeners themselves and more about messing up the ability to nest joists together for efficient shipping.

Granted, it's probably a moot point for this particular condition given that the amplified diaphragm demand is trivial compared to the shear transfer capacity of a tube collector with nominal welding.

Much depends on the situation. OP's situation is CMU shear walls by the sound of it. In my neck of the woods, that would mean that the tube lugs would get welded to small embeds in the CMU which would then become the brittle link in the chain and affect this calculus:

1) If I can make a go of it with joist rollover, I need to do that to stay competitive.

2) If I can make a go of it with a tube lug every other joist space, I need to do that to stay competitive.

For those that subscribe to the unamplified demand / distributed strength argument, do you also disagree with the article I posted above that shear studs in a composite deck should designed for overstrength when delivering diaphragm shear to collectors?

I wouldn't say that I disagree with it, per se but, rather, that it's a bit of a surprise to me and it moves the conservatism needle a bit further over than I usually put it. Some thoughts:

3) I noticed that concrete to concrete shear friction is in the ductile category. That surprised me a bit. I've no doubt that grinding the interface down to a fine powder and then picking up some truss action probably does give you a decent hysteris loop, at least for a few cycles. But then, in a way , it still kind of is a version of unreinforced concrete shear.

4) In many ways, I see well distributed composite studs as being just another version of shear friction, albeit a lesser version between dissimilar materials. Moreover, the very procedure that we use to calculate the flexural strength of composite beams assumes some ductility in the connections between the studs and the slabs. So, in these respects, yes, I was surprised to see composite studs on the list.

5) In many ways, I see well distributed composite studs as being just another version of shear friction, albeit a lesser version between dissimilar materials. Moreover, the very procedure that we use to calculate the flexural strength of composite beams assumes some ductility in the connections between the studs and the slabs. So, in these respects, yes, I was surprised to see composite studs on the list.

6) Obviously my general argument about distributed connections would apply to this as well.

7) The article mentioned that the authors and some of their external associates were using this, more conservative approach despite it not really being codified. I have mixed feeling about that:

a) One the one hand, I celebrate a group going above and beyond what is codified based on their judgment. The rare, "race to the top" as it were.

b) I feel that it harms are profession a bit when you get stuff like this floating around that suggests that what many engineers are doing is incorrect / less correct without that stuff having made it into codes and/or widely accepted design guides yet. In sending out mixed messages to the design and construction community, we undermine ourselves I feel. Collectively, we seem wishy washy in a way that diminishes confidence in what we do. If it were up to me, we'd keep this kind of stuff close to our chests until a consensus is reached and disseminated. That said, I doubt that many contractors are reading Sabelli articles or identifying there reasons for relatively slight differences in connection strategies on EOR drawings.

 

[KootK]

The effect you described above wouldn't be enough to make up for a 300% shortfall in demand--possibly more depending on the SFRS and how it was proportioned by the designer--unless the demand capacity ratio was kept low to begin with, in which case one could argue that the system already was designed considering amplified loads.

That bit about the 300% caught my eye and I'm hoping that you can elaborate. My general thoughts:

1) It sounds to me like you're pitching true capacity design for diaphragm connections rather than the over strength approach. I see an inherent logic in that but, at the same time, to my knowledge we are not enforcing that level of rigor and performance elsewhere where things are non-ductile and the overstrenght approach is used ubiquitously.

2) The whole thing is complicated by the fact that, in multistory buildings, the forces that we use for diaphragm boundary elements are not the same forces that we use to design the VLFRS. As I understand things, the diaphragm forces are higher to reflect higher local accelerations resulting from higher mode effects. So, in this sense, the simple model where we assume a thing to continue attract force until a mechanism forms in the VLFRS is a bit murky. So is using over strength for a force level not actually matching VLFRS mechanism formation in my opinion.

3) Across the board, I suppose that I don't really understand how an over strength approach deals with with the DCR issue that you've tabled in ELF design. To me, it would seem to be a much weaker version of capacity design than is the real thing.

What am I missing in the logic there?

 

[Deker]

Much depends on the situation. OP's situation is CMU shear walls by the sound of it. In my neck of the woods, that would mean that the tube lugs would get welded to small embeds in the CMU which would then become the brittle link in the chain and affect this calculus:

1) If I can make a go of it with joist rollover, I need to do that to stay competitive.

2) If I can make a go of it with a tube lug every other joist space, I need to do that to stay competitive.

Thanks for pointing that out, I overlooked that this was occurring directly above the wall.  In that case, the demand may not be trivial relative to the capacity of the connection.  For anchorage into CMU it's more likely that overstrength should be applied.

b) I feel that it harms are profession a bit when you get stuff like this floating around that suggests that what many engineers are doing is incorrect / less correct without that stuff having made it into codes and/or widely accepted design guides yet. In sending out mixed messages to the design and construction community, we undermine ourselves I feel. Collectively, we seem wishy washy in a way that diminishes confidence in what we do. If it were up to me, we'd keep this kind of stuff close to our chests until a consensus is reached and disseminated. That said, I doubt that many contractors are reading Sabelli articles or identifying there reasons for relatively slight differences in connection strategies on EOR drawings.

I mostly agree, but I would prefer a solution that doesn't result in over-codification.  I think most engineers recognize code as a minimum standard and have no qualms about exceeding that when they have a good rationale for doing so.  You actually touched on one of the main reasons I participate in this forum...aside from how much I learn from others and enjoy reciprocating where I can, I think it's important to hash out issues like the one we're discussing in the hopes of reaching a broader consensus so that we don't have so much variation in how we interpret code requirements / intent.

2) The whole thing is complicated by the fact that, in multistory buildings, the forces that we use for diaphragm boundary elements are not the same forces that we use to design the VLFRS. As I understand things, the diaphragm forces are higher to reflect higher local accelerations resulting from higher mode effects. So, in this sense, the simple model where we assume a thing to continue attract force until a mechanism forms in the VLFRS is a bit murky. So is using over strength for a force level not actually matching VLFRS mechanism formation in my opinion.

Interestingly, we used to design collectors for Ω₀F_x and compare to F_px prior to ASCE 7-10 when the code started requiring that the overstrength factor be applied to the diaphragm forces, as well. A less conservative approach than what the code now requires might be to amplify the first mode response only (reduced by R) and add the elastic response of all higher modes. Another approach is to use the peak response from a NLRH procedure (obviously way beyond what is considered practical for most buildings). In any case, it's clear that ELF diaphragm forces on their own are generally not large enough to preclude inelasticity in diaphragms and collector elements.

For example, if you have twenty connections in line and lose one, the demand in the remaining connections goes up about five percent and, in all likelihood, everything remains just fine. I suspect that the capacity at which any potential unzipping would be arrested in such a system is probably well in excess of an amplified demand. The trouble with this line of reasoning is quantifying it. I wait patiently for some reliability theorist to tell me if the magic number is five, ten, or twenty.

1) It sounds to me like you're pitching true capacity design for diaphragm connections rather than the over strength approach. I see an inherent logic in that but, at the same time, to my knowledge we are not enforcing that level of rigor and performance elsewhere where things are non-ductile and the overstrenght approach is used ubiquitously.

I'm not pitching a true capacity design approach, but I am pitching a design that's informed by capacity design principles.  Under seismic load there is going to be much more overlap between the load and resistance probability curves than is typical for other types of loading.  If you're designing this connection for code level forces and not considering overstrength, there's a high likelihood that you're underestimating the load that the system will experience by 3x to 6x (see the last part of my post in this thread), thereby shifting the load curve to the right. The effect of this shift would far exceed the benefit gained by the effect you described above where the net system strength is greater than the sum of the minimum strength of each individual element. To my knowledge, even the maximum tested values for most elements do not exceed the code nominal strength values by that wide of a margin.

3) Across the board, I suppose that I don't really understand how an over strength approach deals with with the DCR issue that you've tabled in ELF design. To me, it would seem to be a much weaker version of capacity design than is the real thing.

It's not perfect, and that's why it's important for designers to be aware that design decisions made when proportioning the SFRS may impact the amount of force that is delivered to other elements in the load path. A strict capacity based design approach may not be practical, but any element in the seismic load path that is not expected to behave ductility should be designed for some level of force higher than that used to design the SFRS. To me, the idea that I can throw 100 little non-ductile connections in the load path and say that I can design them for the same level of force as my SFRS just because there are so many of them isn't really compatible with that philosophy.

 

[KootK]

Alright, sign me up for TEAM OVER STRENGTH then. I'm substantially convinced.

Taking the drill down a bit further, another place that I see inconsistency is in the calculation procedures normally used to determine the demand in distributed diaphragm connections. Designers will normally make the shear panel assumption and assign all of the diaphragm flexural demand to discrete chords. In reality, you almost always have distributed chords which could amplify the peak shear connection demand by 50%. This, in reference again to the potential unzipping of brittle things.

So, tallying some of this stuff:

1) You've got a real world overstrength factor that might be as much as double the ASCE values.

2) You've got DCR ratio problems only loosely accounted for.

3) You've got diaphragm forces different, and higher than, VLFRS forces in multistory structures.

4) The procedures for calculating distributed shear connection demand are gross approximations.

5) There will be a significant redundancy benefit even if it's difficult to pin down and likely less than 600%.

6) In my experience, many common diaphragm to VLFRS connections are non-ductile. Certainly that is the case with CMU.

To me, the idea that I can throw 100 little non-ductile connections in the load path and say that I can design them for the same level of force as my SFRS just because there are so many of them isn't really compatible with that philosophy.

I agree. But, then, I think that one can be forgiven for not getting overly excited about accuracy given all of these things. In general, and with seismic design especially, I still view structural design as little more than the somewhat intelligent, rough proportioning of things. But, yeah, maintaining a consistent design philosophy is important, even when great precision is not available to be had.

It is still my understanding that, even in high seismic events, failures of these kinds of distributed diaphragm connections in shear are few and far between. What is more often observed is:

1) Failures of collector connections where there is a far greater concentration of diaphragm load and;

2) Failures where out of plane forces on heavy wall systems tear potential VLFRS elements away from the diaphragm in such a way that connections to the diaphragm for tension, rather than shear, is the issue du jour.

 

[Deker]

Glad to have you on board! Couldn't agree more with your comment about proportioning things. If you'll notice, I'm not as hung up on the EXACT force to be used as I am about making an effort to ensure that collectors and shear transfer elements are proportionally stronger than the SRFS.