Shear Friction Reinforcement 1

[spats]

According to ACI 318, shear friction reinforcement needs to be anchored to develop yield strength on both sides of the shear plane. I assume this means development length per Chapter 12. Question is, can you multiply development length by As required/As provided?

 

[JAE]

No - per the shear friction requirements you have to fully develop fy in the bars on either side of the plane. See 11.7.8 for the specific requirement.

This means that As(req'd)/As(prov) cannot be used (see reference to "except where anchorage or development for fy is specifically required..." in section 12.2.5).

 

[spats]

Thanks JAE.

 

[KootK]

I have been meaning to ask this very question.

I have always assumed that the shear friction development length could be prorated by As_req / As. I`m aware of what the code says but I`ve always interpreted that as an under-nuanced statement.

It`s written in the code in the form of Vr = As x u x fy. As such, one would need to develop fy.

If it can also be interpreted as Vr = As x u x fs, then it seems to me that partial development should be acceptable.

Can anyone shed some light on the reason(s) why shear friction bars need to be fully developed? Making stuff up that I don`t know to be true, here are some possibilities:

1) Maybe the shear friction attributable to each bar varies considerably and some bars need to be able to yield in order to redistribute load to others? Sort of a shear friction version of ductility?

2) Perhaps shear friction is a binary phenomenon rather than one which varies gradually? Maybe in the process of shearing across one set of saw-toothed-ish projections, full fy gets engaged?

It has always seemed to me that a number of commonly used details would not work if shear friction development length cannot be prorated. In my neck of the woods, foundation walls do not get poured higher than the underside of ground floor slabs. As such, the only thing keeping the basement walls from caving in is shear friction at the slab / wall interface. I attempt to muster that shear friction by running small diameter dowels up from the exterior of the basement walls and bending them horizontally into the ground floor slab. Generally, there is not enough slab depth available to fully develop the bars. As a belt and suspenders thing, I place longitudinal 15M bars in the bend and hope that I can consider the bars fully developed at the bend like you can with beam stirrups.

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

 

[TXStructural]

Without full development, you cannot rely upon full tension capacity of the reinforcement. The mechanism of a shear friction failure due to failure of the bar development would be catastrophic and sudden. You are relying upon the aggregate interlock with a very small roughness/amplitude, so a very small slip of the bars would create full failure.

While this could be overkill in some instances, the risk is too high to allow it as a rule. Also, since you don't really know what the slip angle would be between the rocks, it would be impossible to compute a more precise answer.

 

[TehMightyEngineer]

I've always thought that one could assume a bar of a lesser yield strength than Fy and thus develop for that lower Fy. Say for example you don't need the full shear friction strength and you run a minimum number of grade 60 bars (420 MPa for the metric crowd) if you don't have the room to develop Fy then I see nothing wrong with assuming you have grade 40 bars and developing for a the assumed lower grade. Obviously your bar is oversized for this but you will get a shorter development length and a lower shear friction strength.

Maine EIT, Civil/Structural.

 

[KootK]

TME: I think that your method would be unacceptable for the same reason that mine may be: lack of a ductile clamping mechanism.

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

 

[KootK]

@TX: I was hoping that this would catch your eye. I know that you do a lot of code committee stuff and I've seen you express strong opinions on shear friction numerous times in the past. If you say that ACI's intention is to preclude prorating based on partial development length, then I will accept that as correct. I'd still like to debate the reasoning though.

Parsing your statements:

1) "Without full development, you cannot rely upon full tension capacity of the reinforcement." Right. But then, if our intention is to prorate, we don't require a clamping force commensurate with fy.

2) "The mechanism of a shear friction failure due to failure of the bar development would be catastrophic and sudden." Agreed. However, we accept brittle, catastrophic failure elsewhere in concrete design. The solution is usually just a higher factor of safety.

3) "You are relying upon the aggregate interlock with a very small roughness/amplitude, so a very small slip of the bars would create full failure." I think that your overall magnitude of slip would be less for bars that are partially utilized. I can think of two factors that would come into play. Firstly, most of the "action" in bar development occurs near where the bar enters the concrete. All other things being equal, a partially utilized bar with partial development length should slip less than a fully yielded bar with full development length. Secondly, based on some euro stuff that I've read, a little bit of concrete (~2 bar dia. deep) effectively spalls away where the rebar enters the concrete. Obviously, this would increase slip.

4) "Also, since you don't really know what the slip angle would be between the rocks, it would be impossible to compute a more precise answer." The first page of the attached sketch is a simplified version of my interpretation of this statement. Do I have your intent right? If so, I would have thought that a probabilistic averaging of the slip angles tributary to each bar would iron this out.

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

 

[KootK]

Over the weekend, I have convinced myself that shear reinforcement does need to be developed for fy in many instances. The reason for this that I can see, however, is that often it is difficult to guarantee that bars won't get yanked on with fy level forces for other reasons having nothing to do with shear friction. See the attached sketch for two examples:

1) The first detail is how I've been attempting to use shear friction to keep basement walls from caving in (I mentioned this detail above). In this case, I can't guarantee that intended / unintended flexure in slab won't stress the dowels to fy and thus potentially rip them out through the underside of the slab.

2) The second detail relates to the horizontal cold joints in shear walls. Any dowels placed near the wall tension chord reinforcement are likely to get strained pretty near to fy. If they're not developed for fy, they may get yanked out of the wall before the wall reaches it's intended capacity. In this case, I think that one needs to do more than develop the dowels. Rather they should be pretty much lapped to the vertical reinforcing in the wall.

I'd love to know others' thought on this.

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

 

[dcarr82775]

What friction is being relied upon for the case with an unroughened surface? Obviously you can't stick a bar in 2", but if you are up to some fair % of the development length, it seems unreasonable to use a reduced strength accounting for the less than full development length. I didn't see a limitation on shear friction for only 60 ksi steel. The development length of 40ksi is less than 60 ksi. So you should be able to do calcs based on 40ksi steel with full development length, even though the steel put in the field is 60ksi. I find it hard to believe that shear friction falls apart if the rod buster in the field uses 60 instead of 40ksi bar. That alone gives a 1/3 reduction for development length that I feel the code allows.

 

[FootNMouth]

I have had to rely on shear friction more times that I wish due to contractor error (i.e. post installing dowels for anchor rod development). In my experience, it is usually impractical to post install reinforcement to a depth that would develop Fy. I typically have provided additional reinforcement until my ratio of:
As,reqd/As,prov ≤ embedment provided/development length. I rationalize this is my head by comparing it to expansion/adhesive anchors in which ductile steel failure hardly ever controls. I use a Φ factor of 0.75. Maybe I should reduce this since I am designing the failure mode to be brittle.

 

[KootK]

@ Dudley: I've taken 0.75 as my phi factor for brittle failure loads. I think that there are precedents for that elsewhere in the code (diagonal tension, over-reinforced beams etc.). I'm curious, is the the situation that you describe using threaded rebar and utilizing shear friction between base plates and concrete piers?

@ Dcarr: for the non-dowel action forms of shear friction, I think that the point would be that pull-out of a 60 ksi bar developed as a 40 ksi bar would still be non-ductile. I agree, however, that for pure dowel action (smooth surfaces), development ought to be less important / unimportant. I find it thoroughly confusing that dowel action is covered under the umbrella of shear friction. For true dowel action, the form of the equation is all wrong and much more attention should be given to edge distances etc. I found some European precast stuff that provides a method for checking dowel action independently of shear friction. I like that much better.

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

 

[TehMightyEngineer]

Maybe I missed something but shear friction shouldn't rely on dowel action at all. The bar is just there to provide a clamping force (As * Fy) and that clamping force is the normal force multiplied by the friction coefficient to get a shear capacity.

Maine EIT, Civil/Structural.

 

[KootK]

I agree with your logic TME. However, the code commentary claims that shear friction for smooth surfaces is primarily based on dowel action.

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

 

[dcarr82775]

For smooth surfaces there isn't anything to clamp, thus dowel action.

 

[KootK]

Another interesting wrinkle is the issue of cohesion. I just learned about this recently.

Shear friction has two components: friction and cohesion. In the Canadian code the two parts are separated and I generally ignore the cohesion because I don't yet trust it. In the US, the cohesion values are built into mu which is why the US mu values are larger than the Canadian ones.

So, in Canada, we can have shear friction without any reinforcement at all! I'm also surprised that:

1) Cohesion and friction can be used together, in concert.
2) Cohesion can be used at all in sections that one would expect to be cracked (e.g. shear walls).

For something that looks like sixth grade physics on the surface, shear friction sure does seem to spark a lot of debate.

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

 

[TehMightyEngineer]

I want to say I saw something in one of the recent ACI Structural Journals on shear friction research... Maybe I can dig through them before the end of the day.

Maine EIT, Civil/Structural.

 

[snowmachine88]

I am looking at this as to how it pertains to retaining walls and slabs built into retaining walls.

@KootK It seems like this is a building heavy crowd, relying heavily on ACI for all concrete matters. However, I am primarily a bridge engineer so my first resource is the AASHTO LRFD Bridge Design Specifications (2013). If you look at Art. 5.8.4.3 AASHTO uses a Mu and cohesion much like the Canadian code, but it has the same Mu values as ACI 318-08. I don't have the 318-14 to know if the Mu values have changed.

I read through other threads on this topic that asked for information on the research of shear friction. ACI doesn't offer many references for the research, but AASHTO list several in the commentary: Loov and Patnaik, 1994; Patnaik 1999, Mattock, 2001; Slapkus and Kahn, 2004. Mattock, Li, and Wang, 1976; Mitchell and Kahn, 2001.

 

[TXStructural]

Current draft of ACI 318-14 still uses the phrase "Reinforcement crossing the shear plane to satisfy ... shall be anchored to develop fy on both sides of the shear plane."

 

[KootK]

@ Snow: thanks for the references. I'll check them out. The Loov work was done down the street from my office and is fascinating. I've always found it odd that his general shear resistance model didn't gain more traction.

@ TX: has there been any committee level talk of ditching that provision? Despite numerous threads on the subject, I still haven't heard a decent explanation for why partial development cannot be used.

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