Automatic subdivision of members with intermediate nodes

[trevdna]

I'd love to see functionality where the solver automatically recognizes the presence of an intermediate node within the length of a member, and internally divides the member into smaller sub-members. With that functionality, an intermediate framing member will behave correctly, without having to subdivide the member by hand. This functionality is really nice to have in commercial solvers.

[JWock82]

Definitely a feature I've been wanting to add. I've been slow to get to it knowing it will take a fair effort, and could cost some computational overhead if I don't do it right.

[maderero]

Hey is there any interest in this still?

Code here: repair_mwe.py

[JWock82]

There definitely is. Is that what your code above does?

[maderero]

Yes it's a minimal working example to generate the example image above. I have not done extensive testing, but from the examples I have run, it is correctly breaking up members at intersections then correctly reassigning the relevant releases, distributed loads, and point loads.

[maderero]

I'm working on incorporating this into Node3D/Member3D/Model3D now. I'll come back with a link shortly.

[maderero]

I've put it up here: https://github.com/maderero/PyNite/tree/repair_model Please try the example under Examples/Braced Frame - Repair Usage.py This will create intersecting vertical and horizontal members. The added repair method takes care of the subdivisions.

I am seeing some warnings coming out of the Visualizer (related to how I am breaking up the distributed loads):

.../PyNite/PyNite/Visualization.py:1173: RuntimeWarning: divide by zero encountered in double_scalars
  step = loadLength/num_steps
.../PyNite/PyNite/Visualization.py:1179: RuntimeWarning: invalid value encountered in double_scalars
  position = (position1[0] + i*step*lineDirCos[0], position1[1] + i*step*lineDirCos[1], position1[2] + i*step*lineDirCos[2])
.../PyNite/PyNite/Visualization.py:1182: RuntimeWarning: invalid value encountered in double_scalars
  length = length1 + (length2 - length1)/loadLength*i*step

Any ideas?

[maderero]

I just pushed a commit d1b03ed that substantially reduces the runtime.

[JWock82]

An update on this feature:

This is my current focus for PyNite. I think I've found a simple way to implement this with minimal code using class inheritance to nest members within other specialized members. I need to modify a few more parts of PyNite, and then run tests and debug, but I expect to have this done very soon.

[JWock82]

Update #2:

I think I've got it implemented. I'm starting to run tests on this feature. Once I'm done testing it I'll roll it out in a new version.

[JWock82]

This feature is now complete as of v0.0.67.