TIC-SERIES MODELING OF ABRASIVE FLOW MACHINED SURFACES

Abstract

Surface texture being the imprint left behind by the machining process, which if analyzed properly can lead to a better understanding of the basic mechanism of surface generation. In those of the machining processes where the tool is fed at a specific rate relative to the work surface, the final roughness is repetitive or periodic in nature and can be easily modeled mathematically. Such surfaces are called deterministic surfaces, On the other hand some surfaces are generated by erosion from random attack of electrical sparks or abrasive grains. Such surfaces cannot be modeled mathematically and are termed as stochastic surfaces. Their random characteristics require statistical modeling. One example of such stochastic surfaces is the surface generated by Abrasive Flow Machining (AFM). AFM generated surfaces are unidirectional and random in nature due to transient media flow conditions. Hence, in the present work, the digitized surface profiles of abrasive flow machined surfaces have been modeled by a stochastic modeling, and analysis technique called data dependent systems. Data Dependent Systems (DDS) methodology provides the difference/differential equation model directly from the data without conjecturing or guessing the correct shape of the autocorrelation or equivalently the form of the model. The form of the DDS models easily yields the Green's function, which can be related to the physical phenomenon of the manufacturing process generating the surface. It DDS technique is used, in the present work, to study the surface-roughness profiles of Brass and Aluminum specimen before AFM, after AFM, and with Magnetically Assisted AFM. AFM was found to improve the surface finish in both Brass and Aluminum as indicated by reduction in the order of ARMA models representing the surface. Application of magnetic field, resulted in improvement in surface finish in case of Brass, while the results were opposite in case of Aluminum. The results were supported by the estimation of Green's function, and the number of dynamic active grains per unit cross-sectional area of extrusion passage. iv