Fill? More Like Explode! by Charles Culp
Filled Surface is one of the most powerful tools in surfacing. Its power lies in the complex algorithms behind the feature. This complexity also means that fill can often produce unpredictable results. Let’s look at fill, and see how better to control it, and make sure that it stays as stable as possible, even after changes to the design.
Filled Surface is a tool that is one of the cornerstones for surface modeling. It works great any time you want to blend existing edges together, where they are five or more open edges. It does, however, still have plenty of applications for "blockheads" as well.
If you have an imported part model with surfaces that are missing or didn’t import correctly, sometimes you can replace them with a filled surface. If a surface imports incorrectly and looks like a crumpled wet rag, you can use the Delete Face command, and select the Delete and Fill option. This will automatically create a filled surface and maintain the solid body--you don’t even need to initiate the fill surface tool.
This same technique can be used for awkward intersections of multiple fillets. Just select the awkward intersecting faces and use Delete Face with the Delete and Fill option, making sure to check the Tangent fill box. Sometimes I will even clean up the edges around the fill first. If you want to see a full demonstration of this, see my 2011 SolidWorks World presentation “Surfacing, Where Do I Even Start?”
These solid modeling examples are important, but most of the power of fill is with surface modeling. Fill is great for blending between five or more open edges, or three or fewer edges. Sometimes I will even use Filled Surface for four edged blends when the existing edges are stretched in one direction, or the resulting topology from a Boundary Surface isn’t what I really wanted. Why does Filled Surface behave differently than Boundary Surface? Because it is based on a fifth-degree polynomial and can have inherently more complex curvature. It also allows for trimmed surface patches. Since all surfaces in NURBS are inherently four-sided, trim will cut back (or trim) the surface so it fills neatly among all the edges you select. This automatic trimming is what makes fill unique and powerful. Clicking the Optimize Surface checkbox with a four (or fewer) edged fill will make it try to fit the patch inside the edges, and not trim a larger surface. This is useful occasionally, but it completely ignores the best part of the tool, so I typically keep this off. If I want to have an optimized patch, I typically use Boundary Surface because I have more granular control of the tangency in each direction. Boundary Surface also allows for the use of connectors--with fill, you just get whatever the system thinks is best.
One technique I use for Filled Surface is to always knit the entire perimeter. One reason is that edges sometimes extend past the intersection between two faces, and knitting the surfaces causes the edge to end at the intersection. Also, a knit can slightly tweak edges to make them line up perfectly (this is part of the knitting tolerance that you can adjust). This means that the curvature can change near where two surfaces intersect. This can lead to changes in the curvature at the corners, and you can sometimes get out-of-control filled surfaces. If you don’t want your surfaces to knit later, you can always separate them again by selecting a set of faces, then knit those as a separate body.
I also (rarely) use sketches as a patch boundary, but only edges of existing surfaces. This allows for tangency control of all edges, as well as allowing for a completely knitted perimeter. If you don’t actually want to use one of the surfaces, you can just use Delete Face to remove it later. These reference surfaces can be used to make sure that a part can be mirrored and maintain tangency across the centerline, or can just be used to make sure the curvature is what you want. You can see in this example that the surface on the right is just for reference; after the Filled Surface is created, I will delete the small rectangular face, and then mirror both the curvy face and the new Filled Surface across the Right Plane.
Another option for Filled Surface is to add Constraint Curves. This allows for constraints internal to the patch. Watch out, as this will probably lead you into trouble. I would not try to tame an out-of-control fill this way. To fix an out-of-control fill, check each corner to see which intersection is not completely curvature continuous. Somewhere, you have too sharp a change in curvature--typically at the intersection of two faces. Trying to add another constraint will only make things worse. Instead, use Constraint Curves sparingly, and with the most amount of freedom possible. Adding a single point will add the least constraint, and that is typically the most I will do. In the example shown, I added a 3D Sketch containing only a construction line, with a sketched point at the end. This allows me to control the height of this dome with a dimension, while maintaining a nice smooth filled surface. You can add simple lines that flow with the existing surfaces, but if you push it too far it may lead to trouble.
Have fun surfacing, and remember the basic rules for all surfacing. Keep your sketches simple, always control your tangency, and don’t let the software control your designs.
Charles Culp is a Project Engineer at Essex Industries, responsible for aerospace projects. He is a top contributor to the SolidWorks Forums, and owner of the SWTuts website.