Peter Brooks
Vice President Mechanical Engineering

Richard Trask
Product Marketing Manager
Bentley Systems Inc.
Exton, Pa.

Take a look at the consumer products in any store and you’ll see eye-catching shapes on items as ordinary as can openers and pencil sharpeners. Products sitting in full view, such as telephones, office equipment, and even computer cases are even more stunning. Chances are good that their designers weren’t toying with 2D CAD systems to fashion these pleasant shapes. They probably engineered them with surface- modeling software.

Although the software is often overshadowed by solid modeling, nothing has replaced surfacing tools when designs must accommodate an ergonomic requirement, or when mathematics defines needed shapes that are neither cylindrical or spherical, such as aerodynamic bodies or the external surfaces on automobiles. Surface modeling also makes better sense when designing thin wall parts and sheet metal enclosures because the software provides tools for creating and tweaking ribs and walls.

Appearance and manufacturability are the greatest challenges for software vendors supplying surface-modeling packages. To hit the sometimes elusive design target of what looks best, a surface-modeling system should have at least five capabilities. Each sets it apart from most other modeling systems.

• The software should be based on nonuniform rational B-splines or Nurbs technology. Nurbs produce the most detailed and mathematically sound curves possible.

The term nonuniform refers to the spacing of the knots or control points for the curve. Knots also provide maximum flexibility when creating a curve that will produce a surface. And rational refers to the ability of the curve to be interpolated through the control points without leaving straight lines at either side of the point. In a solid-modeling system that lack Nurbs, users might start with a block or slab. Changing the surfaces on the block is done with limited tools and rarely attains the needed surface characteristics.

• The software should create complex curves and surfaces, those not easily described by spheres and cylinders or other geometric primitives. Creating a surface out of a B-spline or Nurbs curve lets engineers control the surface by locating or positioning points on a curve. Modifying the position of the control points or adding more control points in tight areas of the curve refines it to a needed effect, and ultimately a new surface. This capability gives users greater flexibility in styling areas with complex shapes such as body panels.

To further complicate matters, a series of curves that define a surface, each with a different number of control points, can result in a twisted surface because the software tries to map each control point to a mate on the adjoining curve. While it is not always possible to control the number of points on a curve, the software should provide a smoothing function to ensure a smooth, accurate, and continuous surface between curves.

Lesser CAM packages that might import surfaces for manufacturing cannot use regular B-splines or Bezier splines. Instead, they calculate a tessellated surface and generate toolpaths based on it. Resulting parts are only close approximations of the design.

• Surfacing software should support reverse engineering. This means it should have features for handling point clouds or point data generated by digitizers, and the capability to turn the information into mathematically accurate surfaces.

Many creative designers prefer to shape wood or clay for the exteriors they are looking for and then turn to a digitizer to get the shapes into the computer. A capable modeling package should be able to turn the points into surfaces with one set of tools and combine the surfaces with additional geometry as it’s needed.

• The models should maintain a history tree or some sort of creation history so that designers unfamiliar with the model could make complex changes without having to start from scratch. Modifying a surface becomes difficult without a history tree that describes its build sequence. Users have to rebuild it or mathematically recreate and recalculate it every time it’s modified. The history tree or feature manager solves the modification problem and rebuilds models faster than other methods.

At least one package has curve and surface evaluators that let users modify surfaces while providing plenty of feedback. For example, evaluation tools give users property information such as the tangency of points on surface edges that have been stitched together. The curve evaluator helps engineers evaluate tangency points of the mating surfaces and provides feedback on the continuity of the surfaces.

• And lastly, a surface modeler should support the creation of parametric and freeform surfaces. Parametrics will allow changing a model by making changes to dimension values. Additional design horsepower comes by combining surface tools with solid-modeling techniques.

For instance, solid-modeling methods provide functions such as extrusions, Booleans operations, surface and corner blending, parametric features, stitching, and capping surfaces into solids. More advanced solids generate helix curves, spirals, mathematical curve formulas, and smooth circular arc interpolations. Surfacing packages bring functions for healing imperfect surfaces, placing patches and planar surfaces, lofting, profile sweeping along multiple traces, and circular sweeping of extrusions. Together, users get a system that provides the best of both modeling worlds.

Design becomes a pointless exercise without thinking about a manufacturing process. A big plus for surface models is that they are just what the manufacturing expert ordered. A production engineer needs only good surfaces on which to generate toolpaths. For example, convert a solid model to an IGES file and send it to a shop, and it receives a surface model.

The process happens every day, but depending on the accuracy of the surfaces relative to the database, translating a file from one format to another often introduces model errors such as gaps between surfaces. On the design side, gaps are less critical. On the manufacturing side they are roadblocks to production. Healing tools in surfacing software fills the gaps. The tools maintain continuity between surface edges and make the surface mathematically accurate. What’s more, upstream applications such as analysis and simulation are also impossible on models with such defects.

Another consideration is the software’s ability to import data from other software packages. Because all systems employ different tolerance values when creating surfaces, importing surface data from a different system requires strong healing tools. That’s because incoming surface models are likely to include errors such as gaps. Look for tools such as continuity control which allows stitching together disconnected surfaces and the ability to translate to and from other systems by means of industry standards such as IGES, STEP, DXF, DWG, and SAT.

Continuity functions are also needed to control how the edges of two surfaces are brought together. When healing, for example, the two should just meet and not extend over one another. If they overlapped, it would produce a step, another manufacturing hazard. Continuity functions bring edges together with tangency points so that the two joined surfaces appear to the NC toolpath generator as continuous geometry.

Older systems had few healing tools and relied on patching to cure gaps. A Coon’s patch, for example, is one such method. It lets users patch the opening with a surface so that an NC toolpath generator would not drop paths into it. Not spotting a gap often meant gouging a part during manufacturing.

Despite advances in modeling systems, the biggest competitor to surface-modeling software remains the traditional paper and pencil. Many industrial design engineers feel the keyboard and mouse stifle their creativity. Only intuitive commands, ease-of-use, and strong free-form capabilities throughout the software will convince this group that productivity gains are still possible.

© 2010 Penton Media, Inc.