Mathematical modelling and optimization of cutting conditions in turning operation on MDN 350 steel with carbide inserts

Hard metals are victorious in offering greater functional life in various critical applications because of their excellent material characteristics. But due to their high hardness, they pose machining problems. Therefore, the current work is intended to identify suitable cutting conditions for machining of hard metal components by carrying out turning experiments.MDN 350 steel is considered as the subject hard metal in the present work, as the literature on machining experiments on the aforementioned metal is limited and there is a wide scope of research for improving its machining performance. The current methodology can be implemented for other hard metals as well. Improvement of tool life, enhancement of rate of production, reduction in cost of production and closeness of surface finish to that of grinding are the major goals of the work. The experimental work is divided into two sets wherein in the first set, the cutting inputs are speed and tool feed rate and the experimental output is flank-wear. Cost of production, tool life and rate of production are the machining performance indicators considered for the first set, which are evaluated based on flank-wear data and empirical formulae. In the second set, rake angle, cutting angle and nose radius of the tool insert are varied and roughness of the machined components is measured. The machining performance indicators of the first set are optimized using graphical method of contour plots. Artificial neural networks technique, which is well known for its versatility to model linear as well as non-linear data, is used to express the surface roughness as a function of tool geometrical variables. Genetic Algorithm, which is an advanced optimization technique known for its intricate search for optimal solutions, is used for optimizing surface roughness with optimal combination of the geometrical parameters. The optimum results of the two sets are confirmed through experimental validation and the deviations are found within 10 %. • Improvement in tool-life by 80 %. • Improvement in production rate by 35.71 %. • Reduction of production cost by 26.08 %. • The developed model is accurate for predicting surface roughness with mean square error of 0.00115 and coefficient of determination of 99 %. • A surface roughness of 3.57 μm is achieved while machining the specimens with optimum tool geometry.