5-axis energy efficient machining of rotary impellers on turn-mill machine tools and development of process prediction models using artificial neural networks

Sustainable manufacturing practices in many industries have been gaining importance in recent years as they are crucial for preventing environmental pollution, reducing energy consumption and eliminating hazardous conditions for workers. This thesis focuses on sustainable manufacturing practices, particularly energy efficiency, during the production of complex parts such as rotary impeller parts with freeform surfaces commonly used in the defense, aerospace and automotive industries. The rotary impeller which is a critical part of high-tech components used in these industries, has complex free-form surfaces on its blades which makes its manufacturing difficult and high energy intense. High-tech 5-axis machining centers or turn-mill machine tools are often used for the machining of such complex surfaces requiring good surface quality and tolerances. In this study, first of all a complete operation planning of rotary impeller has been performed using Siemens NX 9.0 CAD/CAM package for AISI 304 Stainless Steel material and a prototype was manufactured on Mazak i200-ST turn-mill machine tool. The operation plan encompasses all process steps including rough and finish cut, turning, milling and 5-axis surface profiling operations with ball-end milling tools After considering the energy shares of all operation steps, in order to reduce overall consumption, experimental analysis have been carried out on highest energy consuming operations which are rough milling of blade cavities and finish cut of blade surfaces utilizing Design of Experiments(DOE) approach. Input parameters used for experimental analysis are stepover (ae), depth of cut (ap), feed per tooth (fz) and cutting speed (Vc), and output variables observed are Surface Roughness(Ra), Specific Cutting Energy (SEC) and Material Removal rate (MRR). Using the data obtained from machining experiments, relationships between input parameters and output variables have been developed by Response Surface Method (RSM) and Artificial Neural Networks (ANN) for estimation purposes, and ANOVA analysis have been done to understand the influence of input parameters on outputs. Furthermore, the developed ANN models are used for multi-objective optimization of output variables using Genetic Algorithm(GA) method and as a result, Pareto optimal relations are generated. Finally, the results which have been predicted using ANN and optimal machining scenario obtained using GA have been verified by further cutting experiments.