Risa 3D Member Area Load Distribution 1

[MKarr]

Good Afternoon All,

This is the second model I have run into this issue on. I am using member area loads (Parallel to A-B) to distribute a load one way to my floor beams which should then carry the loads to their girders, seems pretty straight forward.

But when I look at the transient loads after running the solution I get the following incorrect results:

-Some of the girders are absorbing the loads
-Loads to floor beams are not evenly distributed, they're nearly even with slight variations in magnitude along the
beams regardless of if a girder is absorbing the load

Some info about the model:

-Floor members are modeled with a y-axis offset to simulate even elevation under composite deck (removing offset
does not affect distribution)
-Load is applied over nodes of individual quadrants as opposed to entire floor (fixed this issue on previous model)
-Floor diaphragm is simulated with crossed rigid links (as H Braces) in each quadrant (removing did not affect load
distribution)
-Issue is occuring during the load case distribution, not as a result of a combination.

I realize I can just change the loads to distributed loads and avoid the issue, but I'd like to be able to use this tool if possible to cut down on time as I'm sure anyone would appreciate.

File is attached

Thanks for your help.

 

[MKarr]

It hit me this morning that the load is applied in the appropriate direction but because my "quadrants" are more trapezoidal shaped rather than square/rectangle that the girders are in the load path. I was able to test this on a simple model. And if I think about it that would be the appropriate distribution in reality given the construction of this structure.

Though that doesn't explain why in the previous model I had this issue when the load was distributed over the entire shape, but then it was correct when I applied the load to the individual quadrants.

 

[JoshPlumSE]

The program will try to "mesh" the area load into smaller triangular pieces. The size of this mesh is based on the joints and members within the area (to some extent), but it is most controlled by the Area Load Mesh size on the Global Model settings. Your area load mesh size is set to 144in^2 which is the default and is great when members are spaced at 10 ft on center. However, when members are more tightly spaced (esp wood joists?) or for any case where you notice the transient distributed loads looking different than you expect you should adjust this mesh size to something smaller.

For a rule of thumb, I usually square 1/10th the joist spacing. So, if my steel joist are spaced at 10 ft on center then 144in^2 works. If my joists are spaced 5 ft on center, then I'm thinking something more like 36 in^2. Effort has been made in recent years to improve the meshing routine and provide an good distribution even when the area load mesh size is too high. So, my rule of thumb is probably a bit more conservative than it needs to be. But, it's still a good starting point.

 

[jayrod12]

Keep in mind that reducing the mesh sizing as Josh indicates makes the solving time increase, sometimes exponentially, so don't reduce it anymore than you need to otherwise you're in for an extremely long solve time.

 

[MKarr]

Thank you both for your input, but that's not exactly what I am seeing in the handful of projects I've started using this feature for. Most projects I have used this feature on have had spaces larger than 10', smaller than 10', or a mix of both spaces. Even spaces as little as 2' O.C. have distributed appropriately with the default setting. The first project I ran into the issue one was a 60'x 30' scoreboard with quadrants equal to 30'x15'. And as said in the initial prompt dropping from the full surface to the quadrants solved that issue.

I'll try adjusting the mesh settings when I find some time to see if that fixed the issue for this current project and report back.

Thanks.

 

[JoshPlumSE]

It's roughly the same issue, just a different aspect of it. It has to do with the way the area load is meshed. When you break it up manually into quadrants you are putting better constraints on how the load is meshing. When it is one large area, the program is meshing on it's own and may not be doing as good of a job.

Another aspect of it is one that I didn't mention. There are two meshing area load meshing routines in RISA-3D. The one we start with tries to constrain the mesh based on all joints and members within the area. This one leads to much better meshing even and is not as sensitive to a large area load mesh size. However, this routine can run into internal errors when geometry is a little off. Beams nearly parallel with the area load mesh edge, members slightly out of plane or such. When that more advanced / accurate mesh fails, the program reverts to a much more simplistic mesh routine. I suspect that your model may be using the more simplistic meshing routine. It is tough to know for sure. There was a version of the program that produced an error message instead of reverting to the simplistic mesh routine. However, now that I no longer work for RISA I don't have access to those older versions of the program. Not sure anyone who's there now would remember the issue. Maybe Caitlin or Debbie would remember it, but they weren't as involved with the issue.

 

[MKarr]

Josh,

I did just now have another project that I am using member area distribution and the transient loads were wonky so I tried your method for changing the mesh and it does appear to make a significant difference. Still has some weird things like loads to e^-17 but much more accurate. Thanks for that.