Jitender Kumar Rai
The rigid body motion of thin walled parts and their elastic-plastic deformations induced during high speed milling are the main root causes of part geometrical and dimensional variabilities; these are governed mainly from the choice of process plan parameters such as; fixture layout design, operation sequence, tool path strategies and the values of cutting variables. Therefore, to avoid failures and achieve better machining results it becomes necessary to judge the validity of a given process plan before going into actual machining. After carefully reviewing previous research works for machined part quality analysis of milled parts, it is apparent that the task of milling process verification and optimization considering the effects of overall machining process parameters viz. fixture layout, operation sequence, tool path and cutting parameters on part quality, is complex, needing the development of a generalized methodology based on scientific principles. It is the purpose of the proposed work to address the problem and develop a sound methodology for the prediction and reduction of errors(resulting from rigid-elastic displacements) induced during machining of thin wall prismatic parts(those needing 2.5 axis machining). In this dissertation, a state of the art milling process plan verification and optimization system called FEM-Mill has been developed. The main novelties of the developed system lies in its two important computational modules namely; (i) the FEM based 3D transient milling simulation environment to predict part errors, and (ii)The rule based expert system for diagnosis and rectification of part error causes(if any). The system also incorporates a newly developed GA (genetic algorithm) model to maximize milling productivity through optimal selection of cutting parameters respecting the machine-tool technological constrains such as, spindle power, cutter safety, cutting parameter limits etc. The aforesaid computation modules along with a feature based process planner and a generalized machining load model(ANN(Artificial Neural Network) based) are successfully developed and implemented using APDL (ANSYS parametric design language) and C++ programming language. The proposed system was extensively tested for a variety of prismatic parts with different degrees of geometrical complexities and various combinations of cutting parameters and fixture assemblies taken from the industrial partners viz., (i) MCM s.p.a, Italy, a medium size machining center manufacturing enterprise and (ii) Quinson s.a., France, aircraft part manufacturer. The obtained transient numerical results namely, the cutting force, workpiece temperature distribution, part deflection and stresses for different part geometries were validated with real field data experimentally obtained during the course of this research work. A good agreement between the numerical and experimental data show the validity of the presented FEM based milling simulation model in handling real field problems. The developed milling simulation model would be an efficient means for analyzing the thermal-structural aspects of machining and their influence on the resulting machined part quality. Thus it allows manufacturing engineers in setting appropriate machining process parameters and obtaining better machining results without needing preliminary cutting trials which generally demand huge investment both in terms of Time and Money. This dissertation deals with a review of relevant literature, design of the modular architecture of FEM-Mill and detailed description of the developed computational methodologies along with demonstrations of the functional capabilities of the system with the help of some industrial test cases.
EPFL2008
