Process Modeling

As with many other manufacturing processes, several computer models have been developed for analysis of powder forging, including finite element analysis and expert design systems. These computer tools aid design of the process, including preform design. Such models were difficult to develop because of the complications introduced by porosity in the material. The existence of porosity and accompanying density change during deformation preclude the use of conventional plasticity equations, which are based on zero volume change. In addition, heat transfer rates in the material depend on the local density level, which changes throughout the preform during the forging process. Recently, material models giving accurate representations of porous material behavior during plastic deformation have been developed. Coupled with modern numerical methods, these material models have led to reliable simulations of the deformation during powder forging (Ref 8). The fracture criterion given in Fig. 6 has also been embedded in the computer code so that fracture predictions can be made.

A reliable simulation code is particularly useful for preform design for powder forging of new parts. Starting with an initial guess of the preform shape, a finite element simulation of the forging of that shape will indicate areas of incomplete densification as well as locations at which fracture may occur. Then modified preform shapes can be attempted in an iterative approach to eventually derive a preform shape that leads to defect-free forged parts. This trial-and-error procedure, however, can be very tedious, especially for complex parts. For this reason, an expert design system was devised to provide an initial guess for the preform design, based on experience with previously forged parts. The part to be forged is first subdivided into regions of primary metal flow, such as back extrusion, forward extrusion, and lateral flow (Fig. 10). This subdivision into regions is done automatically by the program, using geometric reasoning applied to computer-aided design (CAD) data representing the part. Preform design rules accumulated from experience on other parts are then applied to each region. An additional set of rules is applied to evaluate material flow between regions. Next, a decision tree methodology is followed to optimize the preform shapes in each region, based on the objective of minimizing defects, wear, and tool loads. Smoothing functions then blend the preform shapes prescribed for each region. Finally, the design details are determined through limited applications of a finite element simulation code (Ref 9).

Fig. 10 An example of an axisymmetric part. (a) Isometric view. (b) Cross section showing regional subdivisions. (c) Link-node network representation of the part topology for expert system design decisions

0 0

Post a comment