It is difficult, if not impossible, to review all developments in the short space afforded here and within the scope of this document. Therefore attention will be paid to needs, novel approaches, and data rather than on actual procedures. It is also envisioned that actual CAD/CAM/CIM procedures for the aerospace/automotive sectors will be covered in the write-ups of their respective areas.
Current Ability of Alternative Concepts to Provide Customer Wants
Design has been defined in a variety of ways. While some have argued for a global definition of engineering design as the information needed to create, use, and dispose of an artifact, in practice the term design is often used in a very narrow sense. Rather than integrating activities as it should, design unfortunately often segregates such activities as product definition, function derived structural specifications, interaction with customers, and planning in the areas of processing, maintenance, and product disposal. It is in the need for integration of these activities that the design of composite structures differs from that of structures fabricated of other materials. With metals, the technology is itself engineered into the basic stock since the properties in many cases do not change much as the result of the fabrication procedure. With composites, however, properties and performance depend not only on the raw materials, but also critically on how the materials, both individually and as a composite, are processed. This mandates consideration of manufacturing in the design phase. Optimum design with composites can necessitate the design of the material not only from the structural viewpoint, but from the matrix impregnation and infusion aspects as well. An illuminating example of this is in resin transfer molding (RTM). The preform is usually designed from a structural point of view, exemplifying its use as the skeleton for the final part. Often, ease of manufacture will even be considered part of the design profile for the preform. However, it is often forgotten that the same preform must also be designed from the view of infusion, so as to ensure uniform infusion of the resin and proper wet-out of the reinforcement. The same anisotropy that comes from the use of unidirectional and specialty fabrics in order to fulfill strength and stiffness criteria can create directional flow, causing the preform to wet-out primarily in one direction, leaving dry spots in the other. Obviously, there is a critical need for the use of simultaneous or concurrent design for composites.
Traditional designers dealt almost exclusively with the shape of their product. With composites, shape and microstructure are intertwined. Pultrusion, for example, is limited to parts of constant cross-section. Reinforcement microstructures are limited by the process itself, with about 20-25% of the fibers being forced in the axial direction to give the material sufficient strength to withstand frictional forces while being pulled through the die. Filament winding is another process restricted both in the shapes and in the microstructural orientations possible. For composites, microstructure typically refers to the dimensions (length and diameter for fibers), volume fractions, and orientations (with respect to local and global coordinates) of the reinforcement phase in the matrix material. The microstructure of a composite determines to a large degree its performance in terms of mechanical, physical, environmental, economic, and other properties. Most aspects of the microstructure are usually created as the part is being formed into its final shape. It is the inherent scale present in the composite itself that presents both difficulties in analysis and prediction of properties, and also presents opportunities for optimizing performance through changes in the relevant dimensions of the microstructure (Figure 7.6).
Figure 7.6. Inherent Levels of Scale in a Composite -- Opportunities for Design
A series of recent papers (Karbhari and Wilkins 1990, 1991a, 1991b, 1992a) points out the importance of the scale effect at the microstructural level on design, and the interested reader is referred to these for further explanation. It is increasingly becoming possible to build a tailored skeleton in near-net-shape before the actual fabrication of the composite (achieved through the process of resin infusion through the preform). Similar techniques are being developed for MMCs and ceramic-matrix composites (CMCs), thereby making possible the development of "materials- by-design," not merely at a global, structural level, but also at a local, microstructural level. The development of computer based design selection tools is critical to the development of a true CAD base which leaves the designer with the flexibility of choosing materials, configuration, and process, based on the specific needs of the product. The aspects of benchmarking have recently been applied to such a development in the form of a decision support system. The basis of the current approach is in the fact that the problem of selecting an alternative to satisfy multiple criteria is far easier if approached from a deselection, or discrimination, point of view. The object of the exercise then changes from one of selecting the best alternative to one aimed at rejecting alternatives that would not meet the broad limit of specifications. The easiest method of doing this is through the cross-plotting of attributes as in Figure 7.7, and rejecting those concepts that do not fall within specified bounds that signify the range of the demand profile.
Figure 7.7. Cross-Plotting of Attributes
It can clearly be seen that processes 1, 3, and 5 fall outside the acceptable bounds of the demand profile as set through parameters 1 and 2, and hence can be rejected. The deletion of these reduces the further analysis to a smaller number of concepts, making it both easier to handle their selection, as well as making it cheaper in terms of time and money expended on the process.
Discrimination between close calls would then involve more detailed analysis, and in the case of actual design alternatives, it would be feasible to conduct a more thorough investigation of the remaining few concepts. In the case of structures, a full scale finite element analysis could be run on the few remaining analyses, rather than on all the concepts generated by the team, thus providing for a more efficient and economical use of time and funds. Through the use of such decision support system (DSS) tools it is possible to make efficient use of facilities and budgets. A key ingredient in the deselection process is the representation of knowledge such that discrimination becomes an automatic process. The choice of discriminators is thus of great importance and forms the basis for the DSS (Karbhari and Wilkins 1992b). Figure 7.8 depicts an example of the discrimination of primary processes based on the metrics of shape complexity and microstructural control.
Figure 7.8. Deselection of Primary Processes
A similar scheme is also applied to the selection (or rather deselection) of forming processes based on the attributes of geometric complexity and size (Figure 7.9). The graphic pertains to the use of sheet-forming processes after the selection of continuous fiber reinforced thermoplastic prepreg tape (APC-2 in this case) was already made.
Figure 7.9. Relative Applicability of Forming Processes
Thus the primary material form had been selected, and Figure 7.9 is merely the next step in the discrimination phase of design. For the efficient selection of concepts (whether they are related to materials or processes), it is useful to be able to view data in terms of a set of pairwise comparisons. Obviously it is of considerable interest to view the options simultaneously so as to be able to determine the optimum choices based on a number of criteria. A schematic of this is shown in Figure 7.10 for the selection of fibers. Criteria such as shear moduli, tensile moduli, and coefficient of thermal expansion are compared to the fiber strength; whereas strength, cost, and diameter can be compared in relation to each other as against the coefficient of thermal expansion. Such a scheme lends flexibility to the materials-process selection stage, allowing the user or design team the luxury of simultaneously reviewing the performance of a number of options based on a variety of criteria.
Figure 7.10. Deselection as a Part of CAD
Figures 7.11a, 7.11b, 7.11c and 7.11d present the cross-plotting of processes based on five selection criteria. In all these, the letters "TS" stand for "thermoset" and "TP" for "thermoplastic." The criteria of tooling cost, part complexity, repeatability, and level of waste are plotted against a common metric, i.e., production rate. These are among the metrics which are not as readily quantifiable as others, such as cycle time and pressure. However, they are often better discriminators. They also are primarily used as production and/or economic criteria on the basis of which a specific process would be selected. Obviously, the rankings are heuristic and based on judgement, but they do serve as guides for the engineer or design team. For the basis of comparison the processes were ranked on a 1-10 scale with 1 subjectively being the lowest and 10 the highest or best. As an example, under tooling (Figure 7.11a), 1 would represent no need for tooling, whereas 10 would represent the equivalent cost of a tool for injection molding. Similarly a level of 1 in Figure 7.11b represents the complexity of a flat plate, whereas 10 would be representative of an integrated three dimensional structure. The level of waste reuse is an indicator of the efficiency of usage of the material systems in a specific process. A level of 10 represents a process in which almost all waste is reusable, whereas a level of 1 represents a process wherein the product quality would be such that rework and/or a high rejection rate is a normal fear.
Figure 7.11ab. CAD-Based Discrimination (1 & 2)
Figure 7.11cd. CAD Based Discrimination (3 & 4)
Based on these four figures, it is possible for the design team to arrive at conclusions in regard to deselecting (i.e., dropping from consideration) a number of non-viable processes very early in the design stage. This not only saves time and money, but also allows the designer (or design team) to focus on the really important and viable concepts. The visual process also provides a tool whereby it is possible to justify why a concept was dropped or to specify the lack of performance based on specific criteria by the rejected concept. This is not possible in a traditional computer based materials and/or process selection scheme, where deselection is done on the basis of preset and very rigid bounds. This has often led to the exclusion of a concept merely on the basis of its having failed the bound test by a fraction of a percent, even though its performance with regard to other key selection criteria was exemplary. When such a system is linked to a CAD station with facilities for 3-D modeling, feature-based representation, and system representation, true CAD of composites will be possible. The development of materials databases and interchange has been investigated by Sargent (1990) and the interested reader is referred to that publication for further details.
The fabrication of composite structures and products is evolving from labor intensive hand layup methods to automated manufacturing using tape layup, fiber placement, RTM, diaphragm and thermal forming, and other potentially cost effective fabrication procedures. Now, more than ever before, processing costs and problems of repeatability can stall new composite programs right at the profit line. Innovations such as automated methods for laying tape to contoured tools with close conformance have reduced hand labor by about 69%, whereas other developments in automated preform fabrication, as well as the integration of the preform fabrication and molding stages, promise to make an already attractive technology (RTM) even more desirable from the standpoint of economics and productivity. Improvements in process models and controls have resulted in newer methods of monitoring cure conditions, optimizing process cycles, and predicting microstructural changes based on processing conditions, leading to the development of more advanced and reliable composites. An outline of such developments was recently listed as part of a materials forecast (Anon 1991) and hence will not be repeated herein.