Definition
What is a manufacturing feature?
A manufacturing feature is a region of a part's geometry that carries a recognisable manufacturing meaning: a hole, a pocket, a slot, a boss, a chamfer, a thin wall. Where a B-rep model simply describes faces, edges and vertices, a manufacturing feature groups some of those geometric elements into something a process planner or a CAM system can reason about directly — “this is a through-hole,” not “two concentric cylindrical faces and a planar face.”
Turning raw geometry into manufacturing features is the job of feature recognition: interpreting a CAD model from a manufacturing perspective, since the original geometry typically carries no explicit manufacturing semantics of its own.1
Why this matters more than it sounds
A face is a purely geometric object; a feature carries manufacturing implications. Two holes that look identical in a B-rep can require completely different operations depending on their depth, their tolerance, whether they intersect another feature, or which orientation a tool can reach them from. Feature-recognition research has spanned several decades and produced many different approaches to this problem — rule-based, graph-based and volumetric methods among the earlier techniques, and more recently graph neural networks and other machine-learning methods trained directly on B-rep or point-cloud data.1
For manufacturing software, recognising a manufacturing feature is a necessary step before manufacturability or process planning can happen at all: a system cannot plan how to make a pocket if it only sees an unlabelled set of faces.
Frequently asked questions
Is a manufacturing feature the same as a B-rep face?
No. A B-rep face is a raw geometric surface; a manufacturing feature groups faces, edges and their relationships into something with manufacturing meaning, such as a through-hole.
Can software recognise manufacturing features automatically?
Yes, this is called automatic feature recognition (AFR); approaches range from rule-based and graph-based methods to modern machine-learning models trained on CAD geometry.
Why does feature recognition matter for cost?
Because a manufacturability or cycle-time estimate needs to know what has to be made, not just what shape exists.
Notes & references
- Ji, Q. & Marefat, M. M. (1997). Machine Interpretation of CAD Data for Manufacturing Applications. ACM Computing Surveys, 29(3), 264–311. https://doi.org/10.1145/262009.262012 Survey of methods for interpreting CAD geometry into manufacturing features, the foundation of automatic feature recognition.