The designer can evaluate an MWM or any flat mechanical design using the approach of Figure. Using CAD software, the designer defines manufacturing features and builds the geometry and topology of the product shape, including the electrical artwork if it exists, in ACIS, a geometric solid modeler. In addition, the designer inputs tolerances, electrical characteristics (if any), and other product attributes. The system translates the ACIS solid model to STEP geometry and topology and populates the remainder of the STEP-based integrated product model from the entered information.
The first step in variant or generative design evaluation is the generation the OOGT information model for the product. From the STEP-based product model and some additional designer input, the design processor generates the Group Technology (GT) codes of the product design and extracts the additional design details for the OOGT information model, a more abstract product model necessary for design evaluation functions. The system includes procedures for generating the OOGT information for flat mechanical products and a class of layered electromechanical products (MWMs). For mechanical products, the system generates data about the product envelope, the product features, the dimensional and geometric tolerances, and the function, material, stock, and production quantity. For electromechanical products, the system considers the product shape, the electrical classification and function, the electronic components and artwork, the non-soldered and soldered hardware, the electrical characteristics, the fabrication and assembly tolerances, and the materials. Two types of procedures generate the OOGT model. The first translates data directly from the product model (e.g. electrical components). The second type of procedure performs reasoning upon the data in the STEP-based product model or the associated ACIS solid model, which is created from the STEP geometry and topology, in order to calculate design information critical to design evaluation, including feature accessibility, feature volume, thin sections, cross section ratios, and undercuts. For example, Figure shows the solids created to determine if the face of a feature forms an undercut. If both of these solids intersect the ACIS solid model of the product shape, the face forms an undercut in that direction. This undercut direction is an attribute of the feature. Candadai et al. [4] describe in detail these procedures for generating the OOGT information. (Recall that the additional OOGT information is required for design evaluation functions.)
Figure: Schematic representation of undercut detection procedure
The design retrieval procedure exploits the power of Group Technology to search for products that are similar to the product design under consideration. The search retrieves products that are approximately similar to a given design. The designer specifies the attributes of the product that define the search and the required level of similarity for each attribute. These attributes correspond to the those of the GT coding schemes described above. (A product is similar to the given design if it matches or exceeds the similarity threshold for every attribute.) The specification of the attributes and similarity threshold requires no detailed knowledge of the GT coding scheme. The search procedure translates each attribute's similarity threshold to a set or range of values for the corresponding GT code digit. This set depends upon the GT code digit value for the given design and the similarity threshold. For a high threshold, the set might contain only that value; for a low threshold, the set may be a range of values centered around the value for the given design.
The search scans the product databases of the candidate manufacturing partners and reads each GT code. The procedure compares the values of the digits in the GT code to target values for each selected attribute. If all of the examined digits fall into the target ranges, the search retrieves the similar product. (Iyer and Nagi [25] provide additional details into the design retrieval procedures.)
The sort procedure calculates a more accurate similarity measure and ranks the similar products accordingly. In this case, product similarity is a combination of similarity measures for user-specified product attributes. For each attribute, we have defined a procedure that uses the corresponding OOGT information about the products to calculate the attribute similarity. For instance, calculating the similarity measure for holes perpendicular to the top surface requires retrieving data about the diameters of all through holes in both products and determining, for each through hole, the most similar hole in the other product. The process is repeated for blind holes, counterbore holes, and countersink holes. The measures are combined to form the similarity measure for this attribute. Additional details about these procedures are given in Iyer [24].
The sorting procedure combines the attribute similarity measures to calculate the overall similarity measure. for this purpose it uses the Analytic Hierarchy Process to derive weights for the selected attributes from subjective information about the relative importance of product attributes. After deriving an overall similarity measure for each similar product, the sorting procedure orders the set of similar products. The information about the designs, process plans, and performance of the similar products is valuable in estimating the cost and lead time performance of the new product design, in identifying partners that can manufacture the product, and in improving the product design.