Generative feature-based design- z - by-constraints as a means of integration within the manufacturing industry by Mark W. S. Jaques, John Billingsley and Dave Hamson Portsmouth Polytechnic This article examines the development zyxwvutsrq of computer aids within manufacturing industry and proposes an altemative approach to the way we design and the designer‘s role within manufacturing. A feature- based generative design-by-constraints approach is applied, which requires the designer to specify solutions in terms of manufacturing data, which is captured by means of an interactive simulation of machining processes, in which the constraints of equipment, materials and tools zyxwvutsrq are displayed to the designer. The effect of this approach on the integration of all areas within a manufacturing environment is explored, as is the simultaneous design nature of this approach. Introduction The histoty of computer-aided engineer- ing is known to most readersalthough it is not often reviewed, mostly because peo- ple are preoccupied with the develop- ment and investigation of new ideas and solutions to the problem of integrationof computer aids within the manufacturing industty. We feel that it is important, when trying to develop methods of integration, that the reason for the lack of it should be explored. Conventional CAD has developed from draughting packages, which were intended to speed up and improve the drawing process. The potential of ‘digitally stored drawings’ was soon appreciated. The first application was in component classification, an essential enablingpartof group technology, where it was used to store large numbers of drawings in coded form, which would allow cross-referencing and hence the identification of similar components. Similarly, CAM has developed from the useof computer controllers on machines, such as the ’plug board automatic lathe’, to replace the board of sequencing switches and thus greatly reduce set-up times. The next application of digital drawings was in the generation of cutting tool paths for use on the newly developed computer-controlled machine tools. This was achieved by the addition of a cutting tool offset (tool radii) to the boundaryof a drawing. These machines developed into the CNC machines we know today. Asthecost of computingpowerfellstill further,another form of computer aid was developed to perform the costing and scheduling tasks known as ‘computer- aided planning and control’. Accoun- tancy programs were developed to perform wages, costing and ordering functions. This ordering function developed into a separate program known as materials requirement planning (MRP). The program’s function was to order automatically the components needed to construct an assembly. These pro- grams were further developed to perform schedulingfunctionsand became known as manufacturing resources planningpac- kages (MRPII). It can be seen from the descriptions of the origin of these computer aids that each was developed to automate and speed up an existing manual function, and that each was developed indepen- dently and without considerationfor the others. This is an example of a non- systems approach to the problem of computer-integrated manufacture, with the development of low-level sub-systems without first consideringthe system as a whole, leadingto a correspondingincom- patibility of input and output requirements. Approaches to integration Considerable research effort has been expended in an attempt to integrate the design and manufacturing functions [l-41. as this is considered an essential part of flexible manufacturing systems. One approach is to use feature recognition to interrogate a CAD solid model of the design solution in order to identify manufacturing features, which are sets of information related to a part’s descrip- tion, such as holes and keyways. By analysingthegeometric information within the solid model, and combining this with the feature information, manufacturing data are produced. Three fundamental approaches have been identified by Shah and Rogers (21 for associatingfeatures with solid models. These methods are summarised below. zyxw Human-assisted feature recognition Thisapproachhas been used in preparing input for process planning systems, in which a planner could interactivelymark surface types on a two-/three-dimensional 261 Computer-Aided Engineering Journal December 1991