Area counted twice in plastic algorithm

I have come across a case where the plastic algorithm can count an area twice and affect the plastic properties. See below example:

import sectionproperties.pre.library.primitive_sections as ps
from sectionproperties.analysis.section import Section

rect1 = ps.rectangular_section(d=600, b=300)
rect2 = ps.rectangular_section(d=100, b=100).shift_section(100, 450)
geom = rect1 + rect2
geom.create_mesh(100)
geom.plot_geometry()

sec = Section(geom)
sec.plot_mesh()

sec.calculate_geometric_properties()
print(sec.get_area())

sec.calculate_plastic_properties()
print(sec.get_pc())

The section meshes as expected and the correct area is calculated (area = 180000), however I believe the .split_section() method used by the plastic algorithm returns geometries that overlap for the top half of the section (i.e. two boxes are returned from the method and there is no hole in the larger box to accomodate the smaller box). This results in a plastic centroid shifted up the section (y_pc = 306.14, expecting y_pc = 300).

This is not the expected behaviour and it would be great to get a fix to this as this results in the incorrect reporting of plastic properties.

[connorferster]

Working on this…

[connorferster]

By using two analysis methods (shapely + FEM) and because these two geometry processors interpret overlapping geometries differently, I think this problem will be inevitable. Overlapping geometry creates ambiguity (did the user intend for “geometry 1” to take precedence or “geometry 2”?). I think the best way to deal with this issue (for now) is to post a warning and guide the user to a better solution.