Toward an affordable automation scheme of friction stir processing

Abstract

Friction stir processing (FSP), an offshoot from friction stir welding (FSW), is an intricate operation that involves refining the material by a rotating tool. Active control of the resulting size and distribution uniformity of grain structure is desirable. Achieving such a control across the processed area requires real-time control of the process input variables in order to control the pertinent state variables (e.g., temperature, strain, and strain rate) throughout the process. Many active control schemes typically used in friction stir processes (such as position-, speed-, force-, and torque-control schemes) require the utilization of dynamometers to provide feedback to the control loop. Many drawbacks are associated with such utilization including the complexity of the required instrumentation and control systems. Another complexity is the required rigidity of the machine tool needed to perform friction stir processes.

In this work, we advance the notion of eliminating the usage of dynamometers by using the readily available motor current signals from the NC machine tool in the computer numeric control (NC) machines. This approach would drastically reduce the cost of FSP machine retrofitting. Presented in this work are guidelines for the implementation of affordable automation of CNC milling machines to perform friction stir processes. The guidelines are demonstrated by retrofitting a vertical machining center with current transducers to replace the usage of a dynamometer. The current transduces were tapped on the output of the drivers of the spindle and the z-drive motors. A custom LabVIEW software program was developed to control the machining center via direct numeric control mode and to monitor current signals which were in turn, correlated to the generated forces.

To demonstrate the methodology, friction stir processing was performed on magnesium alloy sheets for a wide range of process parameters. The tool rotational speed was varied from 600 RPM to 2000 RPM and the traverse feed from 75 mm/min to 900 mm/min. Current signals were monitored during frictions stir processing and were related to the process forces which were measured using a 4-component dynamometer. Linear relations between thrust force and torques with current signals of the spindle and the z-drive motors were established and the signal to noise ratio for each correlation was investigated.

It was found that the current spindle signals are highly correlated to the process torque where results can be used in a torque control loop without the need for expensive dynamometers. To a lesser extent was the correlation satisfactory between thrust force and z-drive motor signal due to bad signal to noise ratio.