Tool life prediction and management for an integrated tool selection system

Abstract

In machining, it is often difficult to select appropriate tools (tool holder and insert), machining parameters (cutting speed, feed rate and depth of cut) and tool replacement times for all tools due to the wide variety of tooling options and the complexity of the machining operations. Of particular interest is the complex interrelationships between tool selection, cutting data calculation and tool life prediction and control. Numerous techniques and methods of measuring and modelling tool wear, particularly in turning operations were reviewed. The characteristics of these methods were analysed and it was found that most tool wear studies were self-contained without any obvious interface with tool selection. The work presented herein deals with the development of an integrated, off-line tool life control system (TLC). The tool life control system (TLC) predicts tool life for the various turning operations and for a wide variety of workpiece materials. TLC is a closed-loop system combining algorithms with feedback based on direct measurement of flank wear. TLC has been developed using Crystal, which is a rule-based shell and statistical techniques such as multiple regression and the least-squares method. TLC consists of five modules namely, the technical planning of the cutting operation (TPO), tool life prediction (TLP), tool life assessor (TLA), tool life management (TLM) and the tool wear balancing and requirement planning (TRP).The technical planning of the cutting operation (TPO) module contains a procedure to select tools and generate efficient machining parameters (cutting velocity, feed rate and depth of cut) for turning and boring operations. For any selected insert grade, material sub-class, type of cut (finishing, medium-roughing and roughing) and type of cutting fluid, the tool life prediction (TLP) module calculates the theoretical tool life value (T(_sugg)) based on tool life coefficients derived from tool manufacturers' data. For the selected operation, the tool life assessor (TLA) generates a dynamic multiple regression to calculate the approved tool life constants (InC, 1/a, 1/β) based on the real tool life data collected from experiments. These approved constants are used to calculate a modified tool life value (T(_mod)) for the given operation. The stochastic nature of tool life is taken into account, as well as the uncertainty of the available information by introducing a 95% confidence level for tool life. The tool life management module (TLM) studies the variations in tool life data predicted by TLP and TLA and the approved tool life data collected from the shop floor and provides feedback concerning the accuracy of tool life predictions. Finally, the tool life balancing and requirement planning (TRP) methods address the problem of controlling and balancing the wear rate of the cutting edge by the appropriate alteration of cutting conditions so that each one will machine the number of parts that optimize the overall tool changing strategy. Two new tool changing strategies were developed based on minimum production cost, with very encouraging results. Cutting experiments proved that the state of wear and the tool life can be predicted efficiently by the proposed model. The resulting software can be used by machine manufacturers, tool consultants or process planners to achieve the integrated planning and control of tool life as part of the tool selection and cutting data calculation activity.