Core is a separate entity placed in a mould to produce a corresponding cavity - hole or undercut - in the casting. Cores are also used for producing complex shaped pockets and special features (for example, a vertical face without draft) that cannot be produced using a pattern or mould alone. Cores may be dispensable (in sand casting) or permanent (in die casting). In gravity diecasting, either permanent or dispensable cores may be used, usually decided by the core shape - simple or complex, respectively.

A core consists of two portions: the body of the core and one or more extensions (called prints). The body of the core is surrounded by molten metal during casting process. A core has to withstand more heat and for a longer duration than the mould. However, once the casting has cooled, a sand core must easily collapse to facilitate its cleaning out. The prints are necessary to support the core in the mould. They also conduct the heat (and gases produced by a sand core) to the mould.

Cores for sand casting are manufactured by packing specially prepared sand in coreboxes. Core-making processes include oil sand, hot box and cold box, which are suitable for different types of applications. The cavity in a corebox is a negative replica of the corresponding part feature. The corebox is made in two segments (with a parting) to enable removal of the core. Complex cores are prepared by assembling or gluing two or more cores of simpler shapes. The core-related activities: sand preparation, core shooting, coating/treatment and placement in mould, consume significant resources. Thus the number and volume of cores must be minimized to the extent possible, to reduce tooling cost and manufacturing time.


Gravity diecast compressor casing and its core
Cored holes - through and blind, can be automatically identified by geometric reasoning. Undercuts can be identified based on the direction of face normal with respect to the draw direction of mould segments. This is explained in detail next.

A simple yet robust feature recognition methodology can be developed based on Boundary Representation of solid models. Let the part model be completely defined by a set of bounding facets, each facet by three edges and each edge by two vertices. Each facet is also associated with a unit normal vector that points from interior (solid) to exterior (space). The right-hand thumb rule applies to the face normal with respect to the three vertices of the facet. The model conforms to Euler's equation: F+V = E+2, where F, E and V are the number of facets, edges and vertices, respectively.