Someone had asked me about applying the gridshell definition to a closed surface. We have obviously been investigating four-sided surfaces so using what was learned from Dan’s original definition I wrote this one. I had to stop at Sweeping the geodesic curves with the normals because the geometry become extremely expensive to process when trying to offset. I started off lofting a surface from 3 independently parametric circles and using this as the base surface. This way the entire process becomes parametric, from start to finish. The tricky part is rotating the top curve in opposite angles to get intersecting geometries to match.