Master-Slave Connection

[cancmm]

I am modeling a large flat 3D concrete slab using a square mesh, supported at regular intervals in both directions by piles. I've analyzed this model two ways and I can't make sense of the results. In the first analysis, I simply analyzed the slab using pinned connections, which, as expected, gave me large stress concentrations over the support node. To alleviate this, I specified each support node as a master, with the rigidly connected slave nodes being within the boundary of the pile's cross section. When I do this however, the results go completely out of wack. In fact, at the pinned supports (fixed in the vertical direction), the nodes undergo significant vertical displacement (impossible!). How can this be? Am I misinterpreting the use of the master-slave function for this application? Would the slaves not be considered rigidly connected? Is STAAD just screwing with me?

Appreciate any help. Thanks.

 

[structuresguy]

I would not use the master-slave command in this way. It is more intended to contrain groups of nodes for lateral displacement, say to represent a rigid diaphragm, where you have not modeled plate elements. It works well for floor systems, where you have only modeled the floor beams (slab not modeled), to keep all the floor framing retrained to the same displacement. In your case, since you have the plates, they are already constrained to each other anyway.

I know what you mean about stress concentration at the pin supported node. I usually just use a little rational averaging of the highly concentrated stresses to come up with a reasonable number for use in designing the slab reinforcing.

Of course, you could average it out more by adding several nodes for each pile, say dividing the pile perimeter into 8 or 12 segments. That would spread the load out more, but I don't know if the end result would be much different.

 

[cancmm]

I've averaged out the stress concentrations before on previous models, but I've found this method referenced in a few papers so I thought I'd give it a shot. Obviously not the results I was looking for.

An advantage of this supposed method would be to reduce the design moments since the spans would be effectively decreased. In my model, 18" diameter supports are placed at 8' o.c., meaning that the slab's clear span would reduce from 8' to 6.5'. This would help in optimizing the rebar detailing if I could get the model to work.

Regardless, I guess I can't figure out why STAAD allows the master-slave command to superceed the support conditions...

 

[Bill2001]

1) Slave directions at joints may not be supported directions or have displacements imposed.
2) Master and/or slave joints may not be inclined supports.

Hope this helps you.
Bill

 

[RAYC1]

If solid elements are defining the slab, then:
3) Master & Slave directions must not connect to solid elements. Use plates instead.

Use SLAVE FY, if vertical is the only direction controlled; then fix FY & MX & MZ at the master. Use thicker and/or stiffer plates within the pile zones.

 

[JoshPlumSE]

Does STAAD have the ability to model "rigid links"? This would be a beam member that was made out of a weightless material that is very, very stiff.

In my experience, it is almmost always better to use these link members to tie nodes together rather than to use the master-slave.

It's been awhile since I've looked through it. But, in my opinion, the STAAD manual has always been somewhat irresponsible in how they present the Master-Slave node behavior. It's just not very often that this feature can be used without undesirable consequences.

 

[structuresguy]

Staad does not have "rigid links" built in, that I am aware of. I have many times used weightless, but stiff, materials in staad though. I just simply define a new material, that has the stiffness and weight properties that I want. This is what I do most of the time when I have a diaphragm that I want to simulate (not model). I create a zero mass material and apply it to shell elements. I connect the shell elements to the columns. This way, each column is tied together, and I don't end up needing to sub-divide my beams to tie the shell elements to the beams. Since I am not using the shells to tranfers gravity loads to the structure, this method works fine. I then vary the stiffness of the material, and thickness of the elements, to get the approximate diaphragm stiffness that I want. I have found this works very well for floor and roof diaphragms.

 

[structuresguy]

(i wish we could edit posts) I can calibrate my diaphragm stiffness by comparing diaphragm deflection against example problems and deflection equations from the deck manuals.

So, using the method I describe, you could easily create a "rigid link" beam in your model.

And your point on the use of master-slave nodes, and the consequences of such use, is very appropriate. I have many times seen bad results from innapropraite use of this function. When I review the models that our junior engineers make, I always check to see if they are using it, and if so, are they using it correctly. I once got involved in a 5 story moment frame building that was modeled innapropriately using master-slave. The design was done by a senior level engineer at my company, who had just been laid off. I was helping one of our juniors engineers check something while checking shop drawings, as the staad model was giving odd results. As I dug into it, it turned out that all the moment frames were under-designed because of the master-slave command. This ended up being a major issue, as we had to redesign connections, add beam bottom chord bracing to the beams, and about 1/2 the steel was already in place when this happened.