A note on the use of STEP for interfacing design to process planning T. Dereli a, * , H. Filiz b a DepartmentofIndustrialEngineering,UniversityofGaziantep,27310Gaziantep,Turkey b DepartmentofMechanicalEngineering,UniversityofGaziantep,27310Gaziantep,Turkey Abstract This short note demonstrates the use of standard for exchange of product data (STEP) for interfacing design to process planning via a compact featurerecogniser. The methodology used in development of the interface (feature recogniser) makes use of both automatic feature- recognition and feature-based design technologies in order to combine their advantages, and the STEP for the non-problematic and full information exchange. Using the abilities of the STEP, a generic con®guration scheme is developed in which the features are treated as a combination of faces to which geometrical and/or technical information is glued (associated). By this way, the designer is only forced to identify the functional parts of the features when designing the part, which may simplify the component design and result in the effective memory utilisation. The feature recogniser was implemented in C on a PC and tested on a large number of examples with positive results. q 2002 Elsevier Science Ltd. All rights reserved. Keywords: Standard for exchange of product data; Product data exchange; Feature recognition; Feature-based design; Process planning; Adjacency relation- ships 1. Introduction Features are the fundamental elements of a product model. A feature is described as any geometric form that is used in one or more design/manufacturing activities, or it can be an information element representing a region of interest within a product [9]. Feature recognition is de®ned as the grouping of a set of faces on the surface of a part, such that each set corresponds to a feature. The recogni- tion of features involves identi®cation of higher-level features like pockets, holes, etc. from a set of lower level features such as surfaces, edges and vertices. The common methods that have been applied to the recognition of the features are syntactic pattern recognition, state transition diagrams, volume decomposition, set-theoretic or construc- tive solid geometry-based approach, graph-based approach, rule-based approach, neural-network-based approach, trace- based approach, etc. The feature recognition work has several drawbacks as discussed in Ref. [10]. This is probably due to dif®culty of representation of a generic object on a solid modeller, and due to increased complexity of the features on the prismatic parts. One of the obstacles to wide- spread use of the features and the development of better feature recognition systems is the low level of support for feature data exchange via standard for exchange of product data (STEP). On one hand, the application of many systems developed for the feature recognition of 3D has not been considered thoroughly enough to verify its suitability for process planning [5]. Four approaches to using features in CAD/CAM applications have been used in previous work; human-assisted feature recognition, automatic feature recognition, design by features (DBF; sometimes called feature instancing) and feature-based design (FBD). All of the four approaches used within the feature technology have their own bene®ts and hindrances. Therefore, it is pro®table to ®nd or to use a mixed (hybrid) methodology that makes use of the advant- ages of the approaches, while eliminating the drawbacks of individual systems. In this paper, we have presented a feature recognition system, which is one of modules of an in-house process planning system. The input to the system is the STEP ®le created for a part which is modelled on any com- mercially available solid modelling environment. The part model is then translated from the STEP format into an equivalent format (structure) based on boundary repre- sentation (B-Rep) scheme that is accessible and manipul- able in the application environment. Orientation of each face of the part is determined. Relationships between adjacent faces of the part are found based on `concavity', and these relationships are stored in a `relation matrix'. Computer-Aided Design 34 (2002) 1075±1085 COMPUTER-AIDED DESIGN 0010-4485/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S0010-4485(01)00187-7 www.elsevier.com/locate/cad * Corresponding author. Tel.: 190-342-360-1200x2601 ext; fax: 190- 342-360-4383. E-mail address: dereli@gantep.edu.tr (T. Dereli).