Shear Friction: Where and When? 2

[KootK]

Over the years, I've made a rather unexciting hobby out of asking other structural engineers a seemingly simple question: "when do shear friction provisions apply?" I ask because, frankly, I don't know myself -- not with any certainty at least. I get a range of responses, often in combination:

1) Shear friction applies at cold joints.
2) Shear friction applies as an alternate when Vc + Vs can't be made to work. This is dangerous in my mind.
3) Shear friction applies at abrupt changes in cross section, like the interface between the flange and web of a tee beam.
4) Shear friction applies at any assumed future crack. This seems pretty vague to me.

I have come to believe that shear friction must be satisfied at all locations within a member where shear is present. This includes cold joints, abrupt changes in cross section, assumed future cracks, and anywhere that diagonal tension would be checked. Basically, anywhere that a shear diagram is not zero, shear friction needs to be satisfied. Please refer to detail "A" of the attached PDF for an illustration of my thinking on this. I believe that if one imagines a vertical cut through a monolithic concrete beam between stirrups, equilibrium of the resulting free body diagram will demand that a shear resisting mechanism falling under the shear friction umbrella be developed.

Now that I've expressed my heretical view that shear friction needs to be satisfied at all locations in monolithic members, the next logical question becomes: "do I need to check shear friction at all locations?" Every time that I've designed a beam in the past, should I have divided it up into ten segments and checked shear friction at each section? I hope not. In fact, I've come to the conclusion that shear friction need only be checked at cold joints in properly detailed concrete members. Please refer to detail "B" of the attached sketch. I speculate that the compression fields present in most concrete members simulate longitudinal prestress and result in the automatic satisfaction of shear friction demands for monolithic members. One hole in this theory is the very fact that the code provides mu values for monolithic concrete. If shear friction need only be checked at cold joints, why bother with a monolithic value?

So my questions for the forum are:

1) In what situations do you think that shear friction needs to be satisfied?
2) In what situations do you think that shear friction needs to be checked?

Thanks for your help.

KootK

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[pikku14]

I think that the situation illustrated in this slide would be a case to check shear friction with monolithic concrete

 

[KootK]

Yes, yes it would. Thanks Pikku.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[PicoStruc]

Vc is diagonal tension and would not apply to a vertical shear plane

Vc is diagonal cracking but is used all the time with the vertical shear plane dv x bw, check One way shear in slab, in footing, etc... all the structural element usually don't have any shear reinforcement !

I've got 11.2.8.1 in front of me as I type this post. It doesn't mention shear friction anywhere

What i meant is literally if Vf > Vc, by code, you need to add at stirrup. You cannot justify not adding reinforcement by 'proving' using interface shear resistance (with dowel) is enough !!!.


And don't forgot that 'interface shear transfer' suppose a interface... a joint, a crack, etc... so if your interface is not a cold joint but a shear crack.... your member already failed, in a performance point of view.


why risk cracking for some stirrup (that confine concrete and improve global performance of the member !)... but that is my opinion !

 

[bookowski]

It seems like you've thought about this quite a bit so you're unlikely to get a satisfactory answer.

You keep going back to your question #1 about where does shear friction need to be satisfied. I think you're generally correct that along a vertical plane (in your diagram) some type of shear mechanism does need to be satisfied, by basic statics there is no way around that. Whether or not the shear friction provisions of the code are the correct method here is a bit more grey. For one thing the shear friction equations are partially b.s. created to agree with testing "it is therefore necessary to use artificially high values of the coefficient of friction in the shear friction equations so that the calculated strength will be in reasonable agreement with test results". Because the equation doesn't come from mechanics it will be hard to derive a satisfying answer.

A quick attempt to put numbers to it, writing this out as I go so there might be something catastrophically wrong with my logic here but I'll give it a go.
Assume a simply supported beam, leave out phi factors and whether ultimate or not.
M_demand = WL^2/8 (assume k-ft)
V_demand = WL/2 (assume k)
Ast = M/(4d) (using rule of thumb for Ast, takes care of conversion, i.e. sq. in.)

From shear friction:
V_capacity = Ast x 60 (ksi) x 1.4 (kips)
= M/(4d) x 60 x 1.4 (kips)
= WL^2/8 x (1/4d) x 60 x 1.4
= WL/2 x L/4 x (1/4d) x 60 x 1.4
= Vdemand x L/d x (1/16) x 60 x 1.4
= Vdemand x L/d x 5.25

Most likely L/d is always >> 1.0, so V_capacity is >> Vdemand for a simply supported beam.
Or maybe I made a mistake above and it's all nonsense.

 

[KootK]

@Pico: I'm grateful for your participation in this discussion but we are clearly not on the same page. And that's a result of my failure to communicate my ideas effectively. While I don't agree with your latest comments, I don't think that there's anything further to be gained by my attempting to badger you into seeing things my way. I may just be flat out wrong for all I know. We'll just have to agree to disagree on this one. That's a valid outcome from a healthy debate in my mind.

@Bookowski: I have, without hyperbole, been thinking about this for years. I feel that the answers to the questions that I've posed are fundamental to a unified theory of shear, including shear friction. I want that unified theory to a) satisfy my own intellectual curiosity and b) inform my design decisions going forward. The design decisions bit is much less important to me really. We all seem to have an intuitive feel for what needs to be done even if our understanding of "why" is a bit murky.

And you're right, I have been struggling to tease out an answer that I find satisfactory. Your comment has gone a long way to helping in that regard, however, and I thank you for it. It's great to hear at least one other human agree that a shear mechanism akin to shear friction is likely required across the vertical section that I proposed. I was hoping to ultimately achieve a consensus agreement that wherever there is shear, there is a shear friction mechanism at work that is generally a) taken for granted in design and b) non-critical based on some rational explanation (I attempted one). For me, that "discovery" was a rather significant shift in my thinking and I've been hungry for feedback on the concept.

I believe that a major reason that it has been difficult for me to get a KootK-satisfactory answer to my questions is that the questions themselves are pointless from a practical design point of view. My next step will be to try to generate a numerical scenario where Vc + Vs > Vertical section SF. This will probably take the form of a less elegant version of what you've already done Bookowski. Again, thanks for the effort.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[VTPE]

See attached. A simple FBD ives the answer. Also, remember why reinforcing is there. The reinforcing and concrete have the same stress until the concrete cracks, then the reinforcing acts alone. (compatibility analysis anyone?) Once the concrete cracks, there would be no shear friction since the ability to provide a compressive force (required for shear friction) is no longer present.

 

[KootK]

Welcome VTPE.

The concrete is just holding the steel in place? The reinforcing acts alone? You've got to be kidding. Have you somehow missed out on the glacially unfolding revolution that is strut and tie design? From a shear perspective, steel ties are just one half of the truss mechanism that depends, unequivocally, on the presence of concrete struts between those ties. The concrete between stirrups matters a great deal.

The reinforcing and concrete have the same stress until the concrete cracks, then the reinforcing acts alone.

It is strain that the reinforcement and concrete have in common prior to cracking, not the stress.

And you certainly can have shear friction in the presence of shear cracking. Firstly, not all of the beam is in pure shear like the infinitesimal element that you've drawn. There will be a compression block that will do an excellent job of transmitting vertical shear. Secondly, neither dowel action nor aggregate interlock depend on un-cracked concrete. Both mechanisms have been tested and shown to be active post-cracking.

A free body diagram cut between stirrups is every bit as valid as any other FBD. If the concrete is ornamental in your estimation, what is it that you do see transmitting shear between stirrups?

Lastly, conventional shear design is Vc + Vs, not just Vs. Right? Concrete matters?

I think that you've grossly underestimated the importance of concrete "holding the steel in place".

Thanks for posting the sketch VTPE. Taking time out of your day to walk to the scanner = commitment.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[TehMightyEngineer]

I don't believe that shear friction has to be checked where a "traditional" diagonal failure plane applies because of the following:

1) Shear friction to me seems to be inherently lower in stiffness than concrete and steel shear resistance given the usual assumptions such as diagonal cracking and transverse shear reinforcement. Thus, shear will be resisted by the traditional method first and then shear friction second. Essentially, shear friction does not strike me as another limit state for shear, only another way to reinforce for the same limit state.

As an example. Lets say we fillet welded two overlapping steel plates together. We then also took some clamps and clamped the plates together giving a friction force. If we apply a tensile load on each plate then the load is resisted by both the welds and the clamping friction force. However, by inspection we don't check the clamps because it's also welded as they're resisting the same limit state. If the plates couldn't be welded then of course we check the clamping force. However, with the weld the presence of the clamps are redundant and generally not as optimal a solution. Thus, we would either remove the clamps or ignore them.

2) With the exception of strange circumstances I would imagine that that traditional shear resistance design methods would result in a higher shear strength than those provided under shear friction. Therefore, if traditional concrete shear design is satisfactory then you have resisted the shear load. Any added strength from shear friction will not be needed. Further, because shear friction is "less stiff" shear failure would already have had to occur or be occurring before shear friction could be considered engaged. If that were the case and traditional shear design applied but was purposefully not met in favor of shear friction then I would say it was an incorrect design.

3) Shear friction relies on aggregate interlock, compressive friction, and/or dowel action. With the exception of dowel action; after a diagonal shear failure, cracks will have opened up wide enough to prevent the full application of the shear friction method for shear resistance. Thus, only one check or the other should be used, but not both at the same time.

Thus, I feel personally that shear friction only applies when normal shear design assumptions do not apply. To apply it when those assumptions do apply seems to be unnecessary as shear friction is not an additional limit state but rather a "weaker" method of shear reinforcement but less constrained by design assumptions and more applicable to discrete shear failure planes.

Maine EIT, Civil/Structural.

 

[hokie66]

KootK,
Your response to VTPE was excellent. In the diagonal shear provisions, Vc still exists after flexural/shear cracking occurs. But the rest of this thread, and the obvious continuing confusion about shear friction, has reinforced my view that this artificial method is, as bookowski put it "partially b.s.".

 

[KootK]

Nice TME -- thank you. If you read my comments in this thread carefully, you will notice that I've been very careful with my language. I have proposed that:

1) Shear friction must be satisfied everywhere and;
2) Shear friction must be checked only at cold joints.

That difference between where SF should be checked and where it must be satisfied is very important. Perhaps I've been too subtle with that semantic distinction. Like you, I also do not believe that SF needs to be checked wherever diagonal tension is evaluated. I do, however, believe that shear friction needs to be satisfied there.

I believe that your latest comments are almost 100% incorrect TME. Or, more precisely, your comments are almost wholly in opposition to my own views. Since I really do not know that I am in the right here, by definition, I also do not know that you are in the wrong. I have my fingers crossed that you'll take my comments here as they are intended: spirited debate between respected colleagues, not combative assholery.

I thought that your weld plate example was a brilliant device for conveying your ideas. I think that it would be a slightly better analogy if we called it a combination bolted/welded connection. It's more real worldy and the principles are the same, namely deformation compatibility. I'm going to use that analogy below.

Essentially, shear friction does not strike me as another limit state for shear, only another way to reinforce for the same limit state.

If that were the case and traditional shear design applied but was purposefully not met in favor of shear friction then I would say it was an incorrect design.

These two statements seem contradictory to me. I agree with the second one.

1) Shear friction to me seems to be inherently lower in stiffness than concrete and steel shear resistance given the usual assumptions such as diagonal cracking and transverse shear reinforcement. Thus, shear will be resisted by the traditional method first and then shear friction second.

I suspect that you are thinking of the two resisting mechanisms -- diagonal tension (DT) and shear friction (SF) -- as being two alternate mechanisms for addressing the same failure mode (shear). This is analogous to bolts and welds used in the same tension connection. DT and SF are independent failure modes, at the same location, and both need to be satisfied independently. The better analogy would be between bolts and net section rupture in a tension connection. To speak machine, the condition isn't <DT OR SF>, it's <DT AND SF>. As such, the differential stiffnesses of the two mechanisms isn't relevant.

2) With the exception of strange circumstances I would imagine that that traditional shear resistance design methods would result in a higher shear strength than those provided under shear friction.

Hopefully the inverse is true. We check DT because we expect it will govern. We ignore SF because we expect that it will not govern. For this to be true, SF capacity must generally be greater than DT capacity.

With the exception of dowel action; after a diagonal shear failure, cracks will have opened up wide enough to prevent the full application of the shear friction method for shear resistance.

Not so. To quote myself from 2 PM:

you certainly can have shear friction in the presence of shear cracking. Firstly, not all of the beam is in pure shear like the infinitesimal element that you've drawn. There will be a compression block that will do an excellent job of transmitting vertical shear. Secondly, neither dowel action nor aggregate interlock depend on un-cracked concrete. Both mechanisms have been tested and shown to be active post-cracking.

ACI specifically directs designers to use shear friction at locations of real and imagined cracks. That wouldn't make much sense if a crack rendered the method ineffective. When shear cracks from, we don't say that resistance is limited to Vs. Rather, we use Vs + Vc because we acknowledge that shear cracks don't nullify concrete shear resistance. The one glaring exception would be shear within plastic hinges in high seismic zones.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[KootK]

Thanks Hokie. I'll admit that the more I look at shear friction, as a method, the less I love it. Still, it's the law of the land in my particular jungle. I was thumbing through a NZ document that CEL pointed me to last night. I think that I'm starting to understand the non-shear friction view of things. Dowels are treated in true "dowel" fashion. Kind of like mortis and tenon construction in wood. It certainly appeals to my intuition.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[bookowski]

Do a google search for "direct shear failure in concrete beams under impulsive loading" by T. Ross, it's a 200 page pdf. It is focused on dynamic (blast) loading but provides a summary of past research for static loading as well.

 

[TXStructural]

In general, I would apply shear friction only where reinforcement is used to "clamp" surfaces together. This is the case with old + new surfaces, many cold joints, and typical wall-footing joints. Once the reinforcement acts in direct tension (or compression) with a parallel component of force, that is the preferred design scheme. In a beam, strut and tie is the scheme to achieve a safe design (and is the basis for the code-level shear reinforcement design), and shear friction would only apply when a cold joint is allowed to form.

 

[KootK]

@TX: I know that you're a code committee guy. So why the heck is there a mu value for monolithic concrete if shear friction only applies over cold joints? Pikku14 has provided one explanation above.

You've weighed in on the second question of my original post. Care to take a swing at the first? It was most succinctly restated in my supplemental post of [29 Sep 14 11:26].

Your comments reminded me of another issue that I'm interested in. We generally supply shear friction reinforcement in the flexural zone of members subjected to bending. If there's rebar in the compression block as well, does that increase or decrease shear friction capacity? I would expect that it would decrease shear friction capacity since it would shelter the compression block concrete from some of the clamping required for shear friction to develop.


The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[hokie66]

That last question is one of the most ambiguous things about shear friction. Shear friction reinforcement is supposed to provide a clamping force, and there is no way that steel in the compression zone serves to clamp. Thus, those who want to distribute shear friction reinforcement throughout the section are kidding themselves. Distributed reinforcement works for direct tension and can work as dowels, as long as it is not too close to an edge.

 

[TehMightyEngineer]

Kootk: I did miss that distinction and you're entirely right to point it out.

By all means feel free to disagree with me. I definitely tried to make it clear that the above was my opinions based on my engineering judgement alone and with no qualifying evidence to support it. I enjoy a good spirited debate.

These two statements seem contradictory to me. I agree with the second one.

I can see how they could be taken that way. Perhaps limit state isn't the right word. My intent is to say that shear friction is just another way to resist shear, like using bolts instead of welds.

My second sentence was meant to say that traditional shear design per chapter 11.2 and 11.4 must always be checked if applicable but shear friction will not control in those cases. Thus, designing for shear friction without checking the traditional equations of 11.4 would be incorrect.

Below I have conceded this may be wrong.

I suspect that you are thinking of the two resisting mechanisms -- diagonal tension (DT) and shear friction (SF) -- as being two alternate mechanisms for addressing the same failure mode (shear). This is analogous to bolts and welds used in the same tension connection. DT and SF are independent failure modes, at the same location, and both need to be satisfied independently. The better analogy would be between bolts and net section rupture in a tension connection. To speak machine, the condition isn't <DT OR SF>, it's <DT AND SF>. As such, the differential stiffnesses of the two mechanisms isn't relevant.

Hmmmm, I've written about 5 different responses to this and then keep finding a flaw in my argument. I'll concede this only on the basis that I've having trouble refuting it even though it seems wrong to me.

Hopefully the inverse is true. We check DT because we expect it will govern. We ignore SF because we expect that it will not govern. For this to be true, SF capacity must generally be greater than DT capacity.

Whoops, that was indeed backwards. As you've probably noted SF should never control over DT.

In case anyone doubt this:

Per 11.6.4.2, for SF with a diagonal bar (not perpendicular to the failure plane) you get an increase in strength. Per 11.6.7 you're required to resist net shear across the failure plane. Both occur in a DT situation with traditional vertical stirrups. Let's assume these negate each other for a DT situation (as I expect they do).

Lets assume a typical 45 degree crack for this situation and consider only a single crack with a single vertical #4 bar crossing it. Per 11.4.7.5 we can calculate the DT shear resistance for a single #4 bar using equation 11-17.

If we graph the two equations (or just look at them), SF will never control over DT. This is obviously neglecting theta and plain concrete shear strength.

So, perhaps we do always need to consider SF, but because it will inherently never control we just don't bother.

Maine EIT, Civil/Structural.

 

[TehMightyEngineer]

Whoops, three paragraphs were out of order. It should go:

"I can see how they could be taken that way..."

"Below I have conceded this may be wrong."

"My second sentence was meant to..."

Maine EIT, Civil/Structural.

 

[TehMightyEngineer]

As for steel in the compression zone, 11.6.7 and it's commentary appear to address this. It states roughly that flexural tension and compression cancel each other out while net tension requires additional resistance and net compression helps to clamp the surfaces in addition to the shear friction reinforcement. This does seem odd as I would expect that these could not be added together like the code implies it can.

Maine EIT, Civil/Structural.

 

[KootK]

Please review the attached PDF. It is a numerical example of a beam that is adequately reinforced for flexure, passes a diagonal tension shear check, but fails a shear friction investigation. The second page is some MathCAD scribbling for anyone who wants to check my numbers.

While I've been able to come up with numerical example that demonstrates my point, that example is highly contrived. In fact, it took me the better part of an hour to get something to work out the way that I wanted. And, even at that, the proportions of the beam are ridiculous. It's bordering on deep beam territory and, if properly detailed in that context, would likely be self solving for the shear friction check anyhow.

This supports our expectation -- and my hope -- that a vertical shear friction plane is highly unlikely to ever govern the shear design of a properly detailed concrete member.

@Bookowski: Hat's off to you for your on the fly, 9th Grade Algebra-esque, symbolic proof of the same phenomena. Your work has passed my QC review and is confirmed by my clumsy numerical fiddling.


The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[KootK]

@TME: I believe that we're closing in on full agreement here. Thanks for making the effort to respond again.

Regarding 11.6.7., while I agree with your interpretation, it doesn't really address the issue that Hokie and I are concerned about. In a flexural member, some of the rebar often counted on for shear friction winds up being in compression under flexural load. It's hard to imagine how a bar that is in compression can contribute to shear friction clamping. In fact, the force in that bar would resist clamping in my mind. As Hokie has rightly pointed out, a bar experiencing compression could still participate in dowel action if edge distances were appropriate.

Per 11.6.4.2, for SF with a diagonal bar (not perpendicular to the failure plane) you get an increase in strength. Per 11.6.7 you're required to resist net shear across the failure plane. Both occur in a DT situation with traditional vertical stirrups. Let's assume these negate each other for a DT situation (as I expect they do).

It's important to recognize that the situation that you've described here is not shear friction as there is no sliding parallel to the adjacent surfaces. This scenario is just straight up Vc + Vs. Although I agree with your conclusion that vertical plane shear friction is a moot point from a practical perspective, this comparison is not proof of that.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.