Multizone tries to break things down into sweepable/mappable portions, meaning they have the same cross-sectional profile along a path. Because there are no points on your sphere, the software becomes confused and wants to mesh it with tets. So here are three approaches that I see:

  1. Use the meshing in SpaceClaim. Setting up the blocking there is pretty straightforward and there are many videos on YouTube that do similar problem but with more complexity. You can use those as a guide. (see below for simple example)
  2. Use DesignModeler or SpaceClaim to decompose the domain into the sweepable block bodies. Then combine these into a multibody part so they share the nodes between the “bodies”. Here you will basically to project a cube “inside” of your sphere onto your sphere surface. It will look kinda of like a beach ball. Look at examples of O-Grids for a good starting point (there it is basically a 2d version of what we want–again see below).
  3. This is the least elegant and I really should not recommend. But you could use virtual topology in ANSYS Meshing to decompose the faces. The problem here is you don’t have full control of exactly wher the points are on the sphere surface (based on mouse clicks) and changing anything like the geometry would likely require to start over from stratch.