Stochastic Modeling and Analysis of Abrasive Flow Machining

Abstract

Finishing operations in the metal working industry represent a critical and expensive phase of the overall production process. A new process called Abrasive Flow Machining (AFM) promises to provide the accuracy, efficiency, economy, and the possibility of effective automation needed by the manufacturing community. The AFM process is still in its infancy in many respects. The process mechanism, parametric relationships, surface integrity, process control issues have not been effectively addressed. This paper presents preliminary results of an investigation into some aspects of the AFM process performance, surface characterization, and process modeling. The effect of process input parameters (such as media viscosity, extrusion pressure, and number of cycles) on the process performance parameters (metal removal rate and surface finish) are discussed. A stochastic modeling and analysis technique called Data Dependent Systems (DDS) has been used to study AFM generated surface. The Green’s function of the AFM surface profile models provides a “characteristic shape” that is the superimposition of two exponentials. The analysis of autocovariance of the surface profile data also indicates the presence of two real roots. The pseudo-frequencies associated with these two real roots have been linked to the path of the abrasive grains and to the cutting edges of the grain. Furthermore, expressions have been proposed for estimating the abrasive grain wear and the number of grains actively involved in cutting with a view towards developing indicators of media batch life. A brief introduction to the AFM process and related research is also included in this paper.