Toyota has major activities in this area. I was shown demonstrations of body styling CAD, stamping die design, formability analysis of sheet metal parts, CAM of stamping dies, engineering analysis of mechanical parts, process planning of machining, and offline robot programming. I did not see anything comparable to the assembleability analysis that Nissan showed me, how

ever.

clay is also made before the end of this year. A cycle typically takes 40 days, an impossible schedule to maintain if clay were used as the stylists' working medium and as the source of input data for computer models.

Body styling by computer dates to 1981 (Figure 6, top), with a complete end-to-end system working by 1986 or so (Ref 15). Major efforts were made to overcome well-known difficulties with designing and joining surface patches described by earlier theories. Methods of surface generation and curvature evaluation were devised that followed the stylists' methods. Control of surface curvature, its continuity, and its

As at Nissan, computation supports most aspects of vehicle design, including CAD/CAM of body parts, exterior and interior design (interior was not as well developed at Nissan), CAD/CAM of mechanical parts like suspension and power train, structural and aerodynamic Body Styling analysis, laboratory automation, stamping die manufacture, NC programming, and machining process planning. Toyota last made a clay model as primary design data input between 3 and 5 years ago. A (1) making higher quality surfaces good summary of Toyota's computational design work is in Reference 14. Toyota makes about three times as many cars per year as does Nissan. About five new models, plus many minor (3) integrating CAD and CAM redesigns, are in the design system at any time. There are about 200 stylists and 800 body engineers. No data are available on how many mechanical, production, and tooling engineers there are. It appears, however, that Toyota has more in-house people per car design project than Nissan has.

Nissan was a little better than Toyota at presenting the full picture and giving the flavor and comprehensiveness of its long range plans. However, I'm sure that Toyota is ahead in many areas technically. The scientific depth of its work in surface representations and data structures for holding and manipulating design data are two of many examples. The effort reminds one of Nippondenso's commitment to manufacturing equipment excellence as part of manufacturing excellence: it is something you cannot achieve by buying things from vendors.

be the most important factors. Primitive shadowing and rendering of highlight lines were possible in 1983. In the last year, extremely realistic color ren

Higher quality surfaces are smoother, the different sections of the body blend together better, and the final metal realization fits together better. Reduction of leadtime, interestingly, is stated as an explicit objective, something Nissan would not do. Integration of CAD and CAM is a longstanding goal of every car maker. Toyota appears to be several years ahead of other companies in realizing these goals.

The body styling activity takes the first year of the normal 4-year car design cycle. During this year, three or four complete cycles of styling and evaluation may occur. A cycle consists of making three-view sketches, converting them to 3D models and refining these, and making a one-fifth scale clay model by NC machining from the computer data. At least one full size

The color renderings are computed on a parallel computer with 256 transputer elements. Computation takes into account such factors as color, type of paint, weather conditions, and sun angle at various geographic locations. A new car or view angle can be computed in 30 minutes, a new color for the same view angle in 5 minutes. Among the features available that imitates the stylists' old methods is representation of reflections from several fluorescent tube overhead lights.

The styling and rendering system is now used not only to design exteriors but interiors as well. I was shown photos of rough NC-milled clays of dashboards and center consoles (ash tray, shift lever, etc.). The design studio has 65 32-inch diagonal measure flat screen cathode ray tube (CRT) displays (2000x2000 pixels) for the purpose of designing and modifying these surfaces. They run off a UNISYS 2200 mainframe.

The method of converting stylists' sketches to 3D model data demonstrates a widely felt problem, namely, that 3D modelers are too hard to use. Toyota employs specially trained computer technicians who convert the sketches into a first model which the stylist and the technician correct together. The technician interprets the shading in the sketch to obtain an impression of the intended shape, then produces that shape in a surface model. The stylist can view the realistic renderings, an orthographic line drawing, or a cross section. Curves can be modified in ways very similar to those available in Macintosh drawing programs, the most familiar being adjustment of endpoint tangent vector lengths and orientations.

These technicians are obviously rather special people since they must have both an artistic sense and computer skills. They must also provide an important part of the human interface between the stylists and the computer.

Die Face Design and Formability Analysis

Dies cannot be the same shape as the desired final metal part due to the springback of the sheet metal and friction between it and the die face. It used to take about 3 weeks to design a simple die but now with the computer system it takes only 1 week. The functions supported include direct data. transfer from the styling database, addition of shape details for the final part (lightening holes, folds from front to back, locations for fasteners) and details to permit forming (flanges where the die grips the perimeter of the piece), plus formability analyses. These analyses permit the die designer to predict possible forming problems and redesign the stamping process (or occasionally ask the stylist to change the part) to avoid them.

This system is well described in Reference 17 and is credited with shortening die design time by 50%, die manufacturing by 30%, and die tryout by 30%. A major point made by this paper is that the program does not use finite element method (FEM) for stress and formability analyses. Instead, rather basic analyses are used. These include local elongation ratio, speed of deformation in local regions, shape change of grid lines, and other functions that can be computed either from geometry alone or from basic stress analysis. The goal was not a perfect system but one that would help designers find good solutions using methods they could understand and interact with.

Figures 7 and 8 contain a nice example. Here the use of the mean section length ratio is shown. Along a particular feature line, different segments are defined (sections) and their length before and after forming is calculated. The maximum ratio before and after and the rate of change of this ratio along the feature line are cross plotted. Data of this type for 10 past designs were collected and correlated with the die tryout time for each. Excessive tryout times (over 900 hours) lie above the diagonal line, providing designers with rapid feedback on potential problems months in advance. Another feature of this type of analysis is that the designers can put in their own experience, giving them a feeling of ownership and confidence in the program and allowing data to be accumulated for future use or training of new designers.

It is said that in some U.S. car companies methods like this cannot be used effectively because the stylists will not modify their designs. In one company, the stylists report to the chairman of the board whereas the engineers report to the president. In another company, a similar integrated body engineering system is being pursued

but is delayed because many of the component analyses are approximate. Toyota obviously decided not to wait until perfect analyses were available and went ahead to tackle the problems of integrating the existing tools into one system. This decision has put Toyota on a higher plateau, since integration is a new learning opportunity.

Process Planning for Machining "Box-Type" Metal Parts

This system helps designers to choose the necessary machines and tools for making complex parts. An example is a complex aluminum cylinder head with pockets and holes for cam shafts, valves, valve springs, and so on. It is assumed that the part will be made on an existing set of NC machines with a continuous parts conveyor. Parts can circulate on this conveyor and visit any machine in any sequence. Thus transportation capability does not limit process planning.

The part is divided up into regions in two ways: by type of feature (hole, flat) and direction of machining (front, back, perpendicular, oblique). I believe that the software finds the features itself, but I am not sure about this. The machining system is divided into zones containing machines capable of dealing with one or more feature types and directions.

The software makes two types of calculations: finding the right zone for a group of features and planning the cutting conditions for each feature. When several zones are capable of making a feature set, the designer chooses one. He does this apparently without any consideration of workload in the zone from other parts. Each hole feature is classified by a group technology technique using such characteristics as number of steps, tolerance on