Integration of Process Simulation in Machining Parameter Optimization

Abstract

In recent years, simulation tools have proven valuable for the prediction of machining state variables over a wide range of operating parameters. Such simulation packages, however, are seldom an integral part of machining parameter optimization modules. This paper proposes a methodology for incorporating simulation feedback to fine-tune analytic models during the optimization process. Through a limited number of calls to the computationally expensive simulation tools, process parameters may be generated that satisfy the design constraints within the accuracy of the simulation predictions, while providing an efficient balance among parameters arising from the functional form of the optimization model. The following iterative algorithm is presented: (i) a non-linear programming (NLP) optimization technique is used to select process parameters based on closed-form analytical constraint equations relating to critical design requirements, (ii) the simulation is executed using these process parameters, providing predictions of the critical state variables. (iii) Constraint equation parameters are dynamically adapted using the feedback provided by the simulation predictions. This sequence is repeated until local convergence between the simulation and constraint equation predictions has been achieved. A case study in machining parameter optimization for peripheral finish milling operations is developed in which constraints on the allowable form error,Δ and the peripheral surface roughness, Ra, drive the process parameter selection for a cutting operation intended to maximize the material removal rate. Results from twenty machining scenarios are presented, including five workpiece/tool material combinations at four levels of precision. Achieving agreement (within a 5% deviation tolerance) between the simulation and constraint equation predictions required an average of 5 simulation execution cycles (maximum of 8), demonstrating promise that simulation tools can be efficiently incorporated into parameter optimization processes.