ComputerAided_Design_Engineering_amp_Manufactur.pdf

This module gets the necessary inputs from other models of GIFTS and determines the optimum
parameters. These parameters are then passed back to PPIR.
With the introduction of sophisticated and highly expensive CNC machines, the need for the optimal
utilization of these resources is increasingly felt. This requires consideration of the machining economics
in the selection of cutting conditions. It has been realized recently that the optimal process parameter
selection becomes one of the important functions of the process planning activity.
Several optimization models are reported in the literature for turning operations (Balakrishnan and
DeVries, 1982; Prasad et al., 1993). These models can be broadly divided into two categories: (1) unconstrained
optimization models and (2) constrained optimization models.
Unconstrained Optimization Models
Unconstrained optimization has been studied by Brown (1962) and Okushima and Hitomi (1964). The
main disadvantage of unconstrained optimization is that it does not represent a realistic machining
situation because the practical constraints on the machining variables are not considered.
Constrained Optimization Models
Practical constraints acting on the machining parameters are considered in these models. The constrained
models can be classified into single-pass and multipass models. Each of these can further be subdivided
into probabilistic and deterministic models. Probabilistic models consider the uncertainties in the empirical
relations used in model development.
Several single-pass models are reported in the literature. They range from pure graphical solutions to
analytical solutions. The graphical solutions include nomograms (Brewer and Rueda, 1963) and the
performance envelope concept proposed by Crookall (1969).
The multipass models reported by Iwata et al. (1977) and Hati and Rao (1976) come under the category of
probabilistic models. The deterministic models have used optimization techniques such as geometric programming
(Ermer and Kromodihardjo, 1981), sequential quadratic programming (Chua et al., 1991),
dynamic programming (Hayes and Davis, 1979), Powell’s unconstrained method (Yang and Seireg, 1992) and
partial differentiation (Kals and Hijink, 1978). Hinduja et al. (1985) used tool-specific, depth-of-cut-feed
diagrams which represent the possible cutting region to produce easily disposable chips. The selected feed
and depth of cut from these diagrams are tested against various velocity-independent constraints, and then
the corresponding speed is calculated from the “equivalent chip thickness” form of tool-life equation.
Optimization Approach
Some of the points that should be considered while developing an optimization module as part of a
CAPP system are
• Optimization as part of CAPP systems The optimization system should form an integrated module
with CAPP systems.
• Flexibility of mathematical models The mathematical model formulated should be flexible, taking
care of only the necessary constraints and omitting the unnecessary ones. For example, during a
roughing operation the constraints limiting the minimum values of feed and depth of cut can be
ignored, while during a finishing operation the maximum feed, minimum speed, and maximum
depth of cut constraints can be eliminated. This reduces the complexity of the problem, thereby
yielding faster solutions.
• Automatic formulation of problem The optimization system should be capable of automatically
formulating the problem, depending on the type and nature of operation being considered,
without any human intervention.
Inputs for Optimization Module
The optimization module gets the following inputs from the other modules of the CAPP system, as
shown in Figure 5.32.