I have a surface in the shape of an airfoil. I want to offset this surface “inwards” by a constant distance, so I used the Offset Surface component. However, the result is a self-intersecting surface, with some “flaps” at the trailing edge. How can I trim the surface to get rid of these flaps?
offset-airfoil-question.gh (8.1 KB)
This ^ solution is bad, I don’t like it.
As you probably started from 2D section of the airfoil, you totally should create the offset in the 2D phase, and construct the offset surface more reliably…
… or it’s to try with the brep offset method, via script or other plugins
I second what @maje90 says, I would focus on offseting the profile and then producing the 3D via whatever method you wish; sweep, loft, extrude, etc.
Here’s what I came up with. Method 02 shows offsetting the profile curve and then sweeping the profiles along the “wing direction”.
While we can run self intersection on curves, I was not able to figure out a “one node” solution for self intersecting surfaces. Perhaps someone knows of such a component.
Graph Space:
Model Space:
Surprisingly difficult! All efforts so far have failed, including this one:
(connection trouble - all this work for nothing)
I agree with Riccardo that working with 2D edge curves will work better. Here is a first draft - ignore it!
offset-airfoil_2023Sep4b.gh (14.2 KB)
This one works well:
I came across this exact curvature issue
It’s brilliant! I love it. 
This is very interesting, and reminds me of things I’d like to understand more, for some of my own ongoing challenges.
Very interesting approach. I like learning from files like this 
I have similar projects, where I’ll be starting from points or curves, and lofting them, and trying to scale them non-uniformly in a unique way.
Airfoil curves are defined by an explicit equation in the YZ plane, where Z is up and Y is aft:
z+ Ay3 + By2 +Cy +Dy**0.5
This is a cubic curve with an added square root at the end. The cubic is needed to allow for an optional negative radius of curvature near the aft end of the curve. The square root allows for an infinite slope at the front leading edge of the curve. (Many present-day airfoils (mostly military) have an additional 4th degree term Ey**4.)
The parameters A, B, C, and D completely define the shape of the curve. For a wing type surface two curves are needed - one for the top surface and one for the bottom. Both curves will have the same Z value at the infinite slope leading edge point. At the trailing edge the lower curve will have a slightly lower Z value than the upper curve because no one wants to make something that ends with zero thickness.
This is why a standard offset or scale operation for a true airfoil shape doesn’t work - the trailing edge area needs an adjustment to the A - D parameters to change the shape in a way that approximates the desired result, but doesn’t truly match it.
In WW II it became apparent that scale models used in wind tunnel testing weren’t providing sufficiently accurate data. So the “40 by 80” was built for the Ames research Center at Moffett Field, California. The test area is 40 feet high and 80 feet wide.
You can create a brep from the surface then use CreateOffsetBrep
offset-airfoil-question.gh (14.7 KB)
Another way:
The result is untrimmed surface
offset-airfoil-question_.gh (16.1 KB)
Another solution, that more often I do manually on rhino, to trim self-intersecting surfaces:
We can move control points of half of the surface to a different place, that way we have a non-self-intersecting surface, and we can intersect this new surface with copies of itself, so rhino doesn’t go crazy.
It’s enough to have the division of move-notmoved control points far from the intersection event.
