The angle tolerances are only used for 3D STL import while for 2D the boundary reconstruction is handled by the
gridbdr function which isn't vert intelligent. Hopefully there will be more accurate 2D geometry import from CAD files such as DXF in the future as STL is not a good format. Although not ideal, you can quite easily manually correct and split the boundary in this case with something like:
geom = impexp_stl( 'CFD_Outlet_V3_2D.stl', 'import' );
% Step 1 - Identify boundary to split (here 2).
plotgeom(geom,'style',2)
hold on
i_split_b = 2;
% Step 2 - Plot point on boundary and identify point to split (here 15).
b_split = geom.objects{1}.boundaries(i_split_b);
for i=1:size(b_split.edges,1)
plot(b_split.edges(i,1),b_split.edges(i,2),'k.')
text(b_split.edges(i,1),b_split.edges(i,2),num2str(i))
end
i_split_v = 15;
% Step 3 - Copy boundary i_split_b and delete points.
geom.objects{1}.boundaries = [ geom.objects{1}.boundaries, b_split ];
geom.objects{1}.boundaries(end).edges(1:i_split_v-1,:) = [];
geom.objects{1}.boundaries(i_split_b).edges(i_split_v+1:end,:) = [];
figure
plotgeom(geom,'style',2)
Also, note that the latest
FEATool build includes fixes from some of your previous issues, such as locking grid generation, and remembering mesh generation selection in the GUI.
Leon Robers wrote
I have the follwoing issue, that fea tool only recognizes three boundaries for the following geometry, whereas it should be 4. two boundaries are somehow merged. I played around with the Angle tolerances for feature detection in the impexp stl but I cannot solve this issue.
Do you have an idea how the boundaries can be properly recognised?
CFD_Outlet_V3_2D.stlBoundary_Issue.PNG