Optimal number of thermal hotspots selection on motorized milling spindle to predict its thermal deformation

Abstract

Two primary heat-generation sources in motorized machine tool spindles are (i) friction in spindle bearings, and (ii) losses in the motor components. The difference in heat generation and heat dissipation rate from the high-speed motorized milling machine tool spindle cause structural deformation, resulting in part inaccuracy. Several research groups used thermo-mechanical and data-based modelling of high-speed motorized spindles to predict and compensate thermal deformation and improve machining accuracy. In real-time thermal compensation, selection of optimal number of thermal hotspots to place the temperature sensors on the spindle is crucial. The present work proposes an approach to select the optimal number of temperature sensors that can be employed to predict the thermal deformation while maintaining required prediction accuracy. For this purpose, the experiments were conducted to evaluate the thermal heat flux across the spindles at different rotational speeds. Furthermore, a finite element analysis (FEA) was performed to simulate the thermal deformation at the tool center point (TCP). The input parameters considered for predicting the simulated deformation were different temperature sensor data on the motorized spindle and the ambient temperature. This work uses the thermal deformation results obtained from FEA to build an improved second-order polynomial regression model. The model reduces the requirement of temperature sensors from three to one to predict the TCP deflection. Results show that the proposed model accuracy was 86.72% with two temperature sensors and 85.99 % with one temperature sensor. It indicates that one temperature sensor is sufficient to predict the TCP deflection with a compromise of 0.73% prediction accuracy level.